All AI Agents

Claude Code

Anthropic's coding agent for the terminal, IDE and desktop

Claude Code is Anthropic's coding agent. It works on a whole project — from your terminal, and also from a desktop app, VS Code, JetBrains IDEs, the web and Slack, according to Anthropic's documentation — reading files, editing them, running commands and tests, and checking the results before it moves on. Anthropic's pricing page (read 16 October 2026) says Claude Code is included in all paid Claude plans: Pro at $20 a month, or $17 a month on an annual subscription billed $200 up front; Max from $100 a month; and Team, whose standard seats at $20 a month billed annually ($25 monthly) include it, with premium seats at $100 ($125). The free Claude plan does not include it; a Console account on pay-as-you-go API credits, or access through Amazon Bedrock, Google Cloud or Microsoft Foundry, are the other routes. Three features make it workable day to day. A CLAUDE.md project file is read at the start of every session, and Anthropic suggests keeping it under about 200 lines. Checkpoints are saved before each prompt and can be rewound with Esc twice or /rewind — though they do not track changes made by shell commands and are no substitute for Git. And permission modes, switched with Shift+Tab, decide what it may do without asking. The limits: it is slower than autocomplete for one-line edits, usage on subscription plans depends on the plan, and its output still needs a reviewer — Anthropic itself says human review remains crucial for coding agents. Cursor on this site is the editor-first alternative with tab completion, GitHub Copilot sits inside the IDE you already use, and Devin works asynchronously on tickets. Verdict: the strongest choice here for multi-file work in an existing codebase — start on Pro, write a CLAUDE.md, and commit before every large task.

Cursor

AI code editor built on VS Code, priced by usage

Cursor is a code editor built around AI: a fork of VS Code, so extensions and settings carry over, with tab completion, inline edits, chat over your codebase and an agent that makes multi-file changes. Its pricing page (read 16 October 2026) is organised around usage rather than request counts: Free with limited AI use and limited agent requests; Plus at $8 a month with a shared usage pool limited to Grok models; Premium at $30 with four times the Plus usage, cloud agents, automations and Bugbot code review; Super at $100 (15 times Plus) and Ultra at $200 (50 times); Teams from $30 per user with pooled usage; and Enterprise on custom terms. Allowances are expressed only as multiples of Plus, without dollar or request figures, and they refresh weekly; when you run out you buy top-ups or wait, and third-party models are billed separately at API list price, with optional spending caps. No annual prices are shown. Privacy mode, which Cursor says stops your code being used for training by Cursor or its model providers, can be switched on in settings or enforced by a team admin. The limits: usage that is hard to translate into a monthly budget, a Plus plan restricted to one model family, and an editor switch for anyone who works in JetBrains. Claude Code on this site works from the terminal and across editors, GitHub Copilot stays inside the IDE you already use, and Replit Agent builds and hosts whole apps for non-developers. Verdict: the best editor-first AI coding experience here for VS Code users — try Free, and expect Premium to be the tier for daily professional work.

GitHub Copilot

AI pair programmer inside your IDE

GitHub Copilot is the AI coding assistant that lives inside the editor you already use — VS Code, JetBrains and other IDEs, plus the GitHub CLI — completing lines as you type, answering questions in chat and, in agent mode, working through a multi-step task in your repository. What separates it from a terminal agent such as Claude Code is where it sits: it is an assistant in the loop with you, not a worker you hand a ticket to. GitHub's plans page (read 16 October 2026) lists a Free plan at $0 limited to 2,000 inline completions and 50 chat requests a month, Pro at $10 a month, Pro+ at $39 and Max at $100, with Business and Enterprise priced through sales; the Free plan supports the CLI and agent mode. The honest limits: the free quota runs out fast on a working day, chat and agent usage is capped on Free, and suggestion quality depends heavily on how much of the relevant code is open or indexed. Verdict: if your team is on GitHub, Copilot is the lowest-friction way to put completion and chat in every developer's editor; if you want autonomous end-to-end tasks, compare it with Claude Code and Devin on this site.

Devin

Cognition's autonomous AI software engineer

Discover Devin by Cognition, an autonomous AI software engineer starting at $500 per month that plans, codes, tests, and deploys independently.

Free-tier AI coding IDE, formerly Codeium and Windsurf

Originally Codeium's free AI code-completion tool, rebranded to Windsurf in 2025, then acquired by Cognition (maker of Devin) in late 2025 and rebranded again to Devin Desktop in June 2026. Still ships a genuinely free tier alongside paid plans.

Replit Agent

Build and host an app from a plain-English brief

Replit Agent answers one question: can a person without a development environment describe an app and get a working, hosted version back? It runs inside Replit's browser workspace, so planning, code, database, hosting and the live URL all happen in one place — nothing to install, nothing to deploy by hand. That is also its boundary: you are building on Replit's platform, and moving a serious app off it later is work. On the pricing page (read 16 October 2026) there is no standalone free plan; Core is $20 a month, or $18 a month billed annually, and includes $20 of credit towards more powerful models, while Pro is $100 a month ($90 annually) with $100 of model credit. Agent work beyond the included credit is paid for as usage, which is where bills surprise first-time users: a vague brief that needs many rounds of correction costs more than a clear one. Compare it with Cursor if you already code and want an editor, and with Claude Code if you want an agent on your own machine. Verdict: the right first tool for a non-developer who wants a real prototype online this week, and the wrong one for a team with an existing codebase.

Copy.ai

Go-to-market AI platform with a chat entry tier

The first thing to understand about Copy.ai is the gap in its price list. The self-serve Chat plan is $29 a month, or $24 a month billed annually, for up to five seats with unlimited chat words and a choice of OpenAI, Anthropic and Gemini models; the next step, Growth, is $1,000 a month billed as $12,000 a year, with 75 seats and 20,000 workflow credits (copy.ai/prices, read 16 October 2026). That gap tells you what the company is now: a go-to-market automation platform for sales and marketing teams, with the chat tier as the entry door. Where it earns the money is workflows — research an account, draft the sequence, update the CRM — not single pieces of copy, and the credits that meter those workflows are the number to watch. No free plan is listed, though the site offers a trial. If you want a writing assistant for articles, Writesonic or Notion AI on this site are closer fits; if you want prospect-level personalisation, Clay overlaps. Verdict: good for a revenue team that will build and run workflows every week; expensive for everyone else once you leave the Chat tier.

Writesonic

AI search visibility platform with AI articles

Writesonic is no longer best understood as an article generator. Its pricing page (read 16 October 2026) sells an AI search visibility platform: you give it a set of prompts about your brand, it tracks how ChatGPT, Gemini and Google AI Overviews answer them, and its AI articles and site audits exist to close the gaps it finds. Starter is $79 a month billed annually for three tracked AI platforms, 50 prompts, 15 articles, ten site audits and one user; Basic is $199 and Growth $399 on the same basis; Enterprise is quoted and tracks all nine platforms. There is no permanent free tier, but every plan starts with a free trial that needs no card. The limits are the ones that come with the category: prompt counts are small on Starter, AI answers change daily so a single reading proves little, and an AI article still needs an editor before it goes live. If you only want drafting help, Notion AI or Copy.ai are cheaper; if you want to know whether AI assistants are recommending you or a competitor, this is the tool on this site built for that question. Verdict: worth the Starter price for a brand that has already seen traffic shift from search results to AI answers, and premature for one that has not.

Notion AI

AI built into your Notion workspace

Notion AI is not sold as a $10 add-on any more. On Notion's pricing page (read 16 October 2026) the Free and Plus plans carry only a trial of Notion AI, and the full set — Notion Agent, AI Meeting Notes and Enterprise Search in beta — comes with the Business plan at $20 per member a month, with Enterprise on custom pricing; custom agents are metered separately at $10 per 1,000 Notion credits after a free trial. That changes who it is for. If your team already runs its docs, wiki and project databases in Notion, the AI is strong precisely because it works on that content: it summarises a meeting into the page where the agenda lives, fills a database column across many rows, drafts in the doc you are editing and answers questions across the workspace. If your work lives in Google Docs, Slack or a code repository, most of that value disappears and a general assistant is the better buy. The practical limit is the obvious one: it only knows what is in Notion and what you connect to it. Compare Mem on this site for self-organising notes and Glean for search across every work app. Verdict: for a Notion-first team the Business plan is the honest price of the AI; for anyone else it is an expensive way to get a chatbot.

Sudowrite

AI writing partner for fiction

Sudowrite is built for one kind of writer: the novelist, screenwriter or serial-fiction author who wants an AI partner inside the manuscript rather than a chat window beside it. Its tools are shaped around that — describe a scene with more sensory detail, brainstorm the plot, develop characters, match your style, or let the Story Engine draft longer passages — and it keeps your characters and style notes so suggestions stay in your world. It is not a research tool and it is not a marketing tool; ask it for a product description and you are paying for the wrong thing. Pricing (sudowrite.com/pricing, read 16 October 2026) is by monthly credits: Hobby & Student at $10 a month with 225,000 credits, Professional at $22, and Max at $44 with 2,000,000 credits, with a yearly toggle advertising up to 50% off and a free trial that needs no card. Credits go further on short passages than on whole-chapter generation, so a drafting-heavy novelist should budget for Professional or above. On this site the nearest neighbours are Notion AI and Copy.ai, both general-purpose; nothing else here is fiction-first. Verdict: the right choice if you write fiction every week and want the AI to stay in your voice; skip it if your writing is business, academic or SEO.

Perplexity AI

AI answer engine with cited sources

Perplexity AI is an answer engine: ask a question, and it searches the live web, reads the sources and returns a short answer with numbered citations you can click and check. That citation habit is the reason to choose it over a general chatbot for research, and the reason to be careful — a cited answer still needs the citation read, because a summary can flatten what the source actually says. Perplexity's own help centre (read 16 October 2026) sets out the tiers: the Free plan gives practically unlimited basic searches but only three Pro Searches a day and one Research query a month; Pro is $20 a month with weekly Pro Search and monthly Research allowances; Max is $200 a month or $2,000 a year and adds the newest models at higher limits, the Max assistant in the Comet browser and priority support; Enterprise Pro starts at $40 a seat a month, and verified students pay $10 for Education Pro. The limit worth knowing before you pay is that depth, not speed, is the weak point: for a long structured report, Gemini Deep Research is slower but more thorough, and for peer-reviewed literature Elicit and Consensus search papers rather than the open web. Verdict: the default first stop for anything that happened recently and needs a source; the Free plan is enough to find out whether you need Pro.

Elicit

AI research assistant for academic papers

Where Perplexity reads the open web, Elicit reads the literature. You type a research question and it searches a corpus its pricing page puts at more than 138 million papers, returns the relevant studies, summarises each one and lets you chat with the full text; the step that makes it different is the table — pick the papers, name the columns you want extracted (sample size, method, outcome) and it fills the grid, with the source passage behind every cell. Elicit's pricing page (read 16 October 2026) keeps paper search, summaries, full-text chat and Zotero import free on the Basic plan, with limited use of the Research Agent and reports. Pro is $49 per user a month, or $588 billed annually, and is the plan for systematic reviews: a dedicated workflow that can screen 5,000 papers, 20 table columns at a time, custom extractions and API access. Scale is $169 per user a month ($2,028 annually) with five times the usage, figure extraction and live team editing; Enterprise is quoted. The limits: it covers what is in its index, it will not replace reading the papers that matter, and extracted values should be spot-checked against the source before they go into a review. Consensus, also on this site, is the lighter tool when you only need to know which way the evidence leans. Verdict: the best fit here for anyone who has to produce a literature review or evidence table; overkill for a one-off question.

Consensus

Search scientific research

AI that searches 200M+ scientific papers to find consensus on research questions. Get evidence-based answers with direct paper citations.

You.com

Web search and research APIs for AI agents

You.com has changed what it sells. The homepage that used to offer a consumer AI search engine with research, coding and creative modes now leads with web search APIs for agents and LLMs: a Web Search API, a Contents API that returns page text, an Answer API, a Research API in several depths and a Finance Research API, with an MCP server for agent frameworks. If you arrived here looking for a ChatGPT alternative, that product is not what the company is presenting in October 2026, and Perplexity on this site is the closer match. If you are building a research agent of your own, the plans page (read 16 October 2026) is unusually simple: a Free tier with 100 Web Search API queries a day through MCP and no card, then pay as you go at $5 per 1,000 Web Search calls, $1 per 1,000 pages on the Contents API, $5 per 1,000 Answer calls and Research calls from $12 to $1,200 per 1,000 depending on depth, with no seats or platform fee; the page also mentions $100 of free credit, which it equates to 20,000 searches. The limits are the usual ones for an API: you get data, not a finished answer, and quality depends on how you prompt and rank what comes back. Verdict: a good, transparently priced search layer for developers; not a tool for someone who wants to type a question and read an answer.

Gemini Deep Research

Google's research agent that plans, browses and reports

Gemini Deep Research is Google's research agent inside the Gemini app. You give it a question; it turns it into a multi-point research plan, searches and browses the web on its own — Google's overview page says up to hundreds of websites — shows its reasoning as it works, and returns a long, structured report. The same page (read 16 October 2026) says you can upload your own files to include in the research, that reports are also available as audio overviews, and that it works on computer and mobile in 150 countries and more than 45 languages, including for Google Workspace users. On Google's subscriptions page Deep Research appears among the features of the free Gemini plan, with a note that usage limits may apply; the paid Google AI plans raise the Gemini usage limits — Pro to four times the free limit and Ultra higher still. That page shows prices in your local currency, so check the figure for your country. The limits: a long report is only as good as the sources it found, so open the citations before you rely on a figure; it is slower than a quick search by design; and a vague question produces a broad, shallow report, so make the question specific. Perplexity on this site is faster for quick cited answers, Elicit is the specialist for academic papers, and Humata answers from documents you already hold. Verdict: the best tool here for a broad first pass on a market, competitor or topic — start on the free plan and pay only if you hit the limits.

Clay

Data enrichment and outbound automation, metered in credits

Clay is a spreadsheet that goes and gets its own data. Each row is a company or a person; each column calls a provider — email finders, phone data, firmographics, your own API key, or Claygent, an AI agent that researches the row on the web — and the result feeds the next column, until the sheet becomes a personalised outbound list ready for your CRM or sequencer. The thing to understand before you sign up is the two meters. Data credits buy information from Clay's marketplace; actions pay for running enrichments, AI calls and exports; nothing is charged when an enrichment returns no result, and runs on your own API key cost one action and no data credits. On the pricing page (read 16 October 2026) the Free plan carries 500 actions and 100 data credits a month with tables capped at 200 rows; the Launch card shows $167 a month, with a 'starts at' figure of $54 a month for a smaller pack on annual billing, while the page's own FAQ quotes $185 — so treat the entry price as a range and size the credit pack on the page itself; Growth is shown at $446 a month and Enterprise is custom. Compare Amplemarket and 11x on this site if you want the sending and booking done for you rather than a list to work. Verdict: the best data tool here for a team that runs outbound every week; expensive and fiddly for occasional use.

Amplemarket

AI sales platform with sequences, data and Duo

Amplemarket bundles the outbound stack that most teams otherwise assemble from three or four tools: contact data, multichannel sequences across email and LinkedIn, intent signals, and an AI assistant called Duo that drafts and prioritises. Unlike most enterprise sales platforms it publishes an entry price. The pricing page (read 16 October 2026) lists Startup at $600 a month on an annual term for two users and 27,000 contacts, with Multichannel Sequences, AI Intent Signals and Duo Copilot; Growth (four users, 140,000 contacts, adds Duo Voice) and Elite (ten users, 400,000 contacts, adds Duo Inbox) are custom-priced; every plan includes a free trial. Per-user credits are listed too — 13,500 email and 600 phone credits on Startup — which is the real capacity figure to compare against your list. The limits: $600 a month on a year's contract is a serious commitment for a two-person team, the data is only as good as the segment you sell into, and the page does not say which plans include the dialer. If you only need the data, Clay on this site is cheaper to start; if you want an agent that works accounts autonomously, 11x is the comparison. Verdict: a sensible all-in-one for a small, committed outbound team that would rather pay one vendor than wire four together.

11x

AI digital workers for outbound and inbound sales

11x sells software it describes as digital workers rather than tools. Its homepage (read 16 October 2026) names two: Alice, 'email and social agents built to research every account, craft a relevant message, and earn the reply', and Julian, 'voice agents built to qualify and convert inbound demand, and drive expansion revenue'. In practice that means Alice finds and researches prospects and runs the outreach without a rep in the loop, and Julian answers inbound calls and qualifies them. The question a buyer has to settle first is whether they are comfortable with an AI sending the first touch under their brand, because that is the product. Pricing is not published anywhere on the site: every call to action is 'Get a demo', 'Book a Live Demo' or 'Try Voice AI', so expect a sales conversation and an annual contract. The limits are the ones that come with autonomy — reply quality has to be audited, deliverability is your reputation, and an agent that books meetings still needs humans who can take them. If you want to keep people in the loop, Amplemarket gives reps the data and sequences; if you only want the list, Clay does that; both are on this site. Verdict: worth a demo for a company with an outbound motion already proven by humans and the volume to justify automating it; not a first sales hire.

Gong

Revenue intelligence from recorded calls and emails

Gong does not sell to a sales rep; it sells to the people who manage them. It records and transcribes calls, reads the email threads and meetings around each deal, and turns that into coaching moments, deal-risk warnings and a forecast that rests on what customers actually said rather than what reps typed into the CRM. That is why it is in this directory as an agent rather than a dialler: the analysis runs across every conversation without anyone asking. On pricing, Gong's own page (read 16 October 2026) publishes no figures. It states that 'licenses are priced per user', that 'there is a platform fee based on the number of users supported', and that integrating your existing stack is free; instead of a price list it asks for your team size and promises 'a customized proposal'. Budget for both components. The limits a buyer should weigh: recording needs consent in many jurisdictions, adoption depends on reps accepting that every call is reviewed, and the value is in the volume — a team of three will not generate enough conversations to make the patterns meaningful. For outbound execution rather than analysis, see Amplemarket or 11x on this site. Verdict: the standard choice for a sales organisation large enough to coach from data; too much platform for a small team.

Albert AI

Autonomous paid-media optimisation, by quote

Albert AI is what paid-media automation looks like when the human steps out of the daily loop. You connect the ad accounts, set the goals and guardrails, and Albert allocates budget, adjusts bids and audiences and tests creative across channels on its own — the homepage (read 16 October 2026) shows Bing, Instagram, Facebook, YouTube, TikTok and DV360 for programmatic, and a blog excerpt there describes the technology as 'Zoomd-owned Albert AI technology', so Zoomd is the company behind it. There is no price on the site and no demo button either; the only pricing reference is a navigation link, so expect a sales-led engagement sized to your media spend. The honest limits: autonomy at this level needs a brand that already spends enough for an algorithm to learn from, clear conversion tracking so it optimises towards the right event, and a team willing to review rather than operate. For teams that want to stay hands-on, Smartly on this site automates creative and media at scale with humans in charge, and Meta's and Google's own automated campaign types cover most smaller advertisers. Verdict: a candidate for performance teams with significant monthly spend and reliable tracking; not something a small business should wire to its ad account.

AI marketing language for regulated brands

Persado is not a general writing tool, and reading it as one is the quickest way to pick the wrong product. Its homepage (read 16 October 2026) describes an AI creative platform for regulated brands — banks, lenders, insurers and fintechs — that generates, scores and compliance-checks marketing language across email, direct mail, web, social and SMS, and it names CMOs, marketing-operations leaders and CRM owners as the people who buy it. The engine it calls Motivation AI chooses the words and emotional framing most likely to get a response, and the compliance layer is the reason a bank can let it near customer communications at all; the case studies on the site make performance claims, which you should ask them to substantiate on your own data rather than take from a logo wall. There is no price on the site. The calls to action are 'Get a Demo' and 'Start with a Portfolio Audit', and a mention of a 'Self-service Studio tier also available for Optimize' comes with no figure attached. The limits follow from the positioning: it is built for high-volume programmes with controlled approval chains, and a brand without that volume has nothing for it to optimise. For general content drafting Writesonic and Copy.ai on this site are the self-serve options; for paid-media automation see Smartly. Verdict: a serious candidate for a regulated enterprise marketing team that sends at scale; irrelevant to everyone else.

Seventh Sense

Per-recipient email send-time optimisation

Seventh Sense does one narrow thing to an email programme you already run: it looks at when each individual contact has opened and clicked in the past and sends your campaign to that person at the time they are most likely to engage, instead of blasting the whole list at nine o'clock. It plugs into HubSpot or Marketo rather than replacing them, and it adds engagement scoring and frequency control so that fatigued contacts hear from you less. Pricing is by list size, not by seat. The pricing page (read 16 October 2026, HubSpot view) shows a Business plan at $400 a month, or $4,800 a year billed annually, at the default of 25,000 marketing contacts — the page quotes it as $0.192 per contact a year — stair-stepped in 5,000-contact increments, with Enterprise custom-priced from 150,000 contacts and anything above 500,000 by quote; Marketo pricing is not shown separately, and there is a free trial with no card. That price is the main limit: on a small list the lift from timing will not cover $400 a month, and the tool depends on having enough engagement history per contact to learn from. If your email problem is drafting rather than timing, the writing tools on this site are the cheaper fix; if it is campaign strategy, HubSpot's own AI is the place to start. Verdict: worth testing on a large, mature HubSpot or Marketo list; not for a newsletter of a few thousand.

Smartly

Enterprise advertising automation across social, CTV and the open web

Smartly is the automation layer large advertisers put between their brand assets and the ad platforms. Its homepage (read 16 October 2026) organises the product into three suites — Creative, Media and Intelligence — for building ad variants from assets at volume, activating and optimising them across channels, and reporting on what worked; the channels it names are social ('every major social platform'), Connected TV ('200+ streaming services'), the open web through 'all major DSPs', and video. The point is scale: when a campaign needs hundreds of creative variants across markets and formats, the manual design and trafficking time is what Smartly removes. No price appears anywhere on the site; the only commercial actions are 'Get a Demo' and 'Talk to Smartly', so expect an enterprise contract sized to your media spend. The limits are the flip side of the scale: a brand running a handful of campaigns has little for the automation to compress, the platform assumes an in-house or agency team to operate it, and automated creative still needs a brand guardian. Albert AI, also on this site, goes one step further by taking over the optimisation loop itself; for small advertisers, Meta's and Google's built-in automated campaigns do most of this job for free. Verdict: the category standard for performance teams and agencies with enterprise budgets; not a small-business tool.

Surfer SEO

Content optimisation scored against what already ranks

Surfer SEO answers a question writers used to guess at: how thoroughly does this draft cover what the pages already ranking for the keyword cover? Its Content Editor analyses the top results for a target term and gives a live score as you write, with the terms, headings, length and structure that correlate with ranking, plus an outline builder and keyword research; treat the score as a checklist of coverage, not a promise of position, because where a page ends up is decided by far more than the text. The pricing page (read 16 October 2026) shows only yearly-billed monthly rates: Discovery at $49 a month for 120 documents created or optimised, Standard at $99 for 360 documents, Pro at $182, and Peace of Mind at $299 for unlimited documents under a fair-usage policy, with an AI Search Analytics plan at $82 and Enterprise by quote. There is no free plan — the buttons say 'Start for Free' but the page states no trial length — though Surfer offers free tools including an AI detector and an AI visibility audit. The limits: documents are the unit, so a high-volume site burns the quota quickly; the monthly-billing price is not shown in the page text; and optimising to the score can produce pages that satisfy a rubric and lose a reader. Writesonic on this site now tracks visibility in AI answers rather than search results, and a rank tracker, not Surfer, tells you where the page actually lands. Verdict: a sound fit for an SEO team publishing steadily; unnecessary for a site posting a few articles a month.

Zapier AI

Automation across 9,000+ apps, with AI steps and agents

Zapier is the connection layer most teams skip and then wish they had bought first: when something happens in one app, a Zap does the next step in another, and the AI additions mean you can now describe that workflow in plain English, drop an AI step into the middle of it, or hand a whole task to a Zapier Agent. The pricing page (read 16 October 2026) says it connects 9,000+ apps and prices by tasks. The Free plan has 100 tasks a month, two-step Zaps and one seat; Professional starts at $19.99 a month on annual billing or $29.99 month to month for 750 tasks, rising by task tier; Team starts at $69 a month annual for 2,000 tasks; Enterprise is quoted, and annual billing is stated as 33% off. The AI pieces are metered differently: Copilot has a daily message limit on Free and is unlimited on paid plans, AI by Zapier is not included on Free and consumes tasks when it runs, and Agents have basic access on every plan with 400 activities a month on Free and 1,500 on Pro, billed as a separate unit from tasks. That is the limit to understand before you commit: a looping Zap or an AI step on every record can burn through a task tier faster than the headline price suggests. Make, also on this site, is the visual alternative with a lower entry price and credit metering; n8n is the self-hosted one. Verdict: still the default for non-technical teams who want automation to just work; check your monthly task count before choosing a tier.

Make

Visual automation scenarios, metered in credits

Make — formerly Integromat — is what you reach for when a Zapier workflow starts to need an if-branch, a loop over an array or a data transformation between two apps. Its canvas shows the whole scenario as connected modules, so a complex automation is something you can see and debug rather than a list of steps, and that visibility is the real reason technical users prefer it. Pricing is by credits, where each module action normally costs one. Make's own comparison page (read 16 October 2026) lists a Free plan with 1,000 credits a month, Core at $9 a month, Pro at $16 and Team at $29 — each starting at 10,000 credits a month — with Enterprise custom; the page does not say whether those figures assume annual billing, and Make's pricing page itself did not render the figures in text when we read it, so confirm on the page for your billing choice. The same page gives the app count as both 3,000+ and 3,500+. The limits: the learning curve is steeper than Zapier's, AI and complex modules can consume more than one credit per run, and 1,000 free credits disappear quickly on anything scheduled. Zapier on this site is the simpler choice for linear automations; n8n is the self-hosted option when you want to own the server. Verdict: the best value in this category for anyone comfortable with a visual builder; skip it if you want automations to be invisible.

Mem

AI notes that organise themselves

Mem is for people who refuse to file. You write notes, capture meetings and jot tasks, and the AI does the organising — linking related notes, surfacing the right one when you ask, tracking tasks it finds in your writing — so there are no folders or tags to maintain. The paid tiers add routines (scheduled briefings, research and reminders) and connections to sources such as Gmail, Slack or Todoist, which is where it stops being a notebook and starts being an assistant over your notes. The pricing page (read 16 October 2026) shows monthly figures only: Free with 25 messages and 25 new notes a month and three hours of Voice Mode; Plus at $9 a month with 50 of each; Pro at $29 with 100 messages and notes, one custom routine and one connection; then Pro 2× at $49, 5× at $99, 10× at $149 and 20× at $199, each roughly doubling the allowances, with extra usage at $50 a block. Those caps are the limit to read twice: 25 notes a month on Free is a trial, not a workspace, and the per-message metering means a heavy user ends up on a tier that costs more than a general assistant. Notion AI on this site is the choice when your notes need to live in structured databases with a team; Glean is enterprise search across apps rather than a personal notebook. Verdict: a good fit for a solo knowledge worker who wants capture without maintenance; count your monthly notes before you pick a tier.

Reclaim AI

AI time blocking for teams

AI automatically schedules tasks, habits, and meetings in your calendar based on priority and availability. Protects deep work time.

Otter AI

AI meeting notes and transcription

Real-time transcription, meeting notes, and action items automatically generated from Zoom, Meet, and Teams calls. Searchable meeting archive.

Julius AI

Plain-English data analysis in an AI workspace

Julius started as an AI data analyst — upload a spreadsheet, ask a question in plain English, get the analysis and a chart — and its pricing page (read 16 October 2026) now presents it as a broader AI workspace that also makes presentations, reports, websites, images and video. Analysis is still the core. It works on spreadsheets, documents and data files; connects to Google Drive, OneDrive and SharePoint on every plan and to Snowflake, BigQuery and Postgres on paid plans; and its FAQ adds MySQL and SQL Server and describes building a semantic layer of your schema so it can query warehouses with thousands of tables. The plans: Free at $0 with daily credits and Julius's lighter models; Plus at $20 a month, or $16 billed annually, with 2,000 credits a month, frontier models, 16 GB of RAM for analysis and seven-day file storage; Pro at $45, or $37 annually, with 5,000 credits, 32 GB of RAM and unlimited parallel agents; Business at $450, or $375 annually, with 60,000 credits, shared workspaces and up to 50 team members; and Enterprise by contact, with single sign-on, audit logs and permanent file storage. The page says Julius is SOC 2 Type II compliant. The limits: files are kept for seven days below Enterprise, so it is not a place to store data; credits are shared between analysis and the creative tools, so image and video work eats into your analysis budget; and like any AI analyst it can pick the wrong method confidently, so check the code and the numbers before a result goes into a decision. Equals on this site is the spreadsheet-first alternative, Polymer builds shareable dashboards, and Zams — formerly Obviously AI — now sells sales-operations automation rather than prediction. Verdict: the most capable plain-English data analyst here for individuals and small teams — start on Free with a real file, and choose Plus or Pro by how often you analyse.

Zams (formerly Obviously AI)

AI workers and automation across your business apps

If you arrived here looking for Obviously AI's no-code prediction tool, start with the fact that it no longer exists as a live product. The old site (read 16 October 2026) carries a banner — 'Obviously AI is now Zams. This site is an archive of how it operated through 2025' — and the company now trades as Zams, with the footer on zams.com reading 'Obviously AI, Inc. (d/b/a Zams)'. What Zams sells is different: an 'AI workforce' that connects email, calendar, CRM, drive and databases so a team can search, track metrics and automate tasks in plain English, plus named AI workers for sales teams. The Zams pricing page lists an Automation Platform at Basic $125 a month for one user with 1,000 agent tasks, Pro at $1,208 a month with 50,000 agent tasks, Team at $1,417 a month for 25 users, and Enterprise by contact; among the AI workers only Evan, a meeting-briefing agent at $50 a month, is priced, with Iris, Nico, Nova and Atlas marked coming soon. Annual billing is advertised at 20% off without the figures shown, and there is no free tier — the FAQ offers a guided trial on request. The limits: a product in mid-transition, no public predictive-modelling offer any more, and a price list that reads as enterprise from the second tier. For analysing a dataset in plain English, Julius on this site is the direct tool; for dashboards, Polymer. Verdict: evaluate Zams as a sales-ops automation platform on a demo, not as the spreadsheet-prediction tool the old name promised.

Equals

AI-powered spreadsheets

Spreadsheet that connects to your data warehouse with AI analysis built in. Ask questions, get SQL written automatically, visualise instantly.

Dashboards from spreadsheets and marketing accounts

Polymer turns a spreadsheet or a connected account into a dashboard without a BI project. You upload a CSV or XLS, or connect Google Sheets, Shopify, Facebook Ads, Google Analytics, Google Ads or Salesforce, and it builds visualisations you can filter, embed and share, with an AI chat layer for asking questions of the data; how often it re-syncs, how many accounts per connector you can attach and how many AI responses you get are what separate the tiers. One warning before you buy: Polymer's pricing page (polymersearch.com/pricing, read 16 October 2026) shows more than one plan table and they disagree. The most detailed one lists Basic at $10 a month or $5 a month billed yearly for file uploads and Google Sheets with manual syncing and no AI chat; Starter at $50 a month ($25 yearly) with all connectors and daily syncing; Pro at $100 ($50 yearly) with hourly syncing; Teams at $250 ($125 yearly); and Enterprise by contact — while another table on the same page lists Starter at $20 and Pro at $40. Read the table attached to the button you press. There is no permanent free plan; the page mentions a free trial of 7 days in one place and 14 in another, and says students and educators can use it free. A second warning: polymerhq.io is a different company selling data security for AI and SaaS tools, so make sure you are on polymersearch.com. For asking analytical questions of a dataset rather than building a dashboard, Julius on this site is the closer tool. Verdict: a quick, inexpensive way for a marketing or e-commerce team to get live dashboards; confirm the plan table before paying.

Intercom Fin

AI customer-service agent priced per outcome

Fin is priced by the thing you actually want: a customer conversation that ends without a human. Intercom's pricing page (read 16 October 2026) lists Fin 'from $0.99 per Fin outcome' on every seat plan and as a standalone product, and defines an outcome as a conversation in which the customer confirms the issue is resolved, does not ask for more help after Fin responds, or in which Fin completes a workflow including a handoff — charged once per conversation however many questions it answered. Around that sit the human seats: Essential at $29, Advanced at $85 and Expert at $132 per seat a month as displayed, plus a Copilot add-on for agents at $29 a month billed annually or $35 monthly after ten free Copilot conversations per agent. Fin itself answers across email, live chat and phone, can take actions in external systems and hands off to the inbox when it should. The limits to budget for: the per-outcome meter scales with your volume, so a busy support desk pays far more than the seat price suggests; answer quality depends on the help-centre content you give it; and a handoff counts as an outcome, so measure deflection yourself rather than taking the dashboard's word. Tidio on this site is the small-business equivalent with a free plan; Zendesk AI meters by automated resolution on top of per-agent seats. Verdict: the most transparent pricing model in the category, and the right fit for a SaaS or e-commerce team already on Intercom; price your monthly conversation volume before you switch for it.

Tidio

Live chat and the Lyro AI agent for small business

Tidio starts free and stays that way for a surprising number of small shops. The pricing page (read 16 October 2026) lists a Free plan at $0 with 50 billable conversations a month and ten seats, plus 50 Lyro AI conversations as a lifetime allowance rather than a monthly one — so it is a trial of the AI inside a genuinely free live-chat tool. From there Starter is $24.17 a month for 100 billable conversations, Growth starts at $49.17 for 250, Plus starts at $300 plus usage, and Premium is by contact, with an annual toggle offering two months free. Lyro, the AI agent that answers customer questions from your help content, is sold as an add-on 'from $32.50 a month' for 50 Lyro conversations, scaling to 1,000 before you need a custom plan. What you get is live chat, chatbots and email automation, with native Shopify actions from the Growth plan up, which is why it keeps appearing in small-business stacks. The limits: billable conversations are the real meter and a busy store will outgrow Starter quickly; Lyro's quality tracks the quality of your help pages; and once you are paying for both the plan and the Lyro add-on, compare the total with a per-outcome tool. Intercom Fin on this site is that comparison for SaaS teams; Zendesk AI for larger support desks. Verdict: the best starting point here for a small e-commerce or service business — begin on Free, watch the conversation count, and upgrade only when it bites.

Zendesk AI

Enterprise support with AI agents metered per resolution

Zendesk's AI comes with two meters, and both have to be in the budget. The first is the seat: the pricing page (read 16 October 2026) lists Suite Team at $55 per agent a month paid yearly or $69 monthly, Suite Professional at $115 yearly or $149 monthly, and Suite Enterprise with Copilot by sales conversation. The second is the automated resolution — a request the AI agent resolves without escalating to a human. Suite Team includes five of those per agent a month and Professional ten; beyond that they cost $1.50 each on a committed volume or $2.00 pay as you go, and the FAQ states you pay only for requests resolved without a human. On top sits Copilot for human agents at $50 per agent a month paid yearly, and features such as suggested first replies are listed as further add-ons without a price. What the AI does is the full enterprise set: ticket triage and intent detection, AI agents answering customers, agent assist and quality review. The limits: the two meters make the real cost hard to see on day one, the included resolutions are small, and the platform assumes a support organisation with process and admin time. There is a 14-day trial on Professional features, and a startups programme offering six months for up to 50 agents. For a small shop, Tidio on this site does the job on a free plan; for a per-outcome model without per-agent seats on the AI side, see Intercom Fin. Verdict: the right platform for an established support team that already needs Zendesk's ticketing; expensive as an AI purchase on its own.

Image and video generation priced by GPU time

Midjourney is the image generator people choose for the look of the picture rather than for convenience. You write a prompt, it returns a set of images you can vary and upscale, and on the higher plans it animates them into short video. Its plan comparison page (read 16 October 2026) lists four subscriptions: Basic at $10 a month with 3.3 hours (200 minutes) of Fast GPU time, Standard at $30 with 15 hours plus unlimited image generation in Relax mode, Pro at $60 with 30 hours, and Mega at $120 with 60 hours. Committing to a year takes 20% off — $8, $24, $48 and $96 a month, paid upfront — and extra Fast time costs $4 an hour on every plan. Two plan rules matter more than the prices. Stealth mode, which keeps your images and videos private, exists only on Pro and Mega; and the page says a company making more than $1,000,000 in gross revenue a year must buy Pro or Mega. Unlimited Relax video is also limited to those two plans, and HD video starts at Standard. The limits: GPU time, not image count, is the meter, so heavy iteration on Basic runs out quickly; work made below Pro is not private; and exact text, logos and layouts are harder to control than in a design tool. Canva AI on this site is the faster route to a finished social graphic, and Runway is the pick when the output is video first. Verdict: the strongest choice here for image quality. Most working creatives should start on Standard for the unlimited Relax images, and a business over the revenue line should budget for Pro from day one.

Runway

AI video generation with a professional model line-up

Runway is the video generator to compare everything else against, mostly because of what its pricing page (read 16 October 2026) lists on each plan. The Free plan is free for ever with 125 one-time credits that do not expire and 5 GB of storage; Standard is shown at $12 a month billed annually with $15 struck through, for 625 credits a month and access to Gen-4.5, Gen-4 Turbo, Aleph 2.0 and third-party models including Kling 3.0 and Seedance; Pro is $28 billed annually ($35 struck through) with 2,250 credits and Topaz AI upscaling; Max is $76 billed annually ($95 struck through) with 9,500 credits and 500 GB of storage. Credits are the unit — every generation spends some, more for longer or higher-resolution clips — and on Standard and Pro unused credits do not roll over, while Max allows one month to carry across. Around the generators sit the editing tools: motion brush, image-to-video, text-to-video and the rest of the toolkit that made it a professional standard. The limits are the category's: short clips, not continuous scenes; a usable shot often takes several regenerations, which is your real cost per clip; and the page shows the discounted annual rate more prominently than the monthly one, so check what you will actually be billed. PixVerse on this site is the lighter, social-first alternative; Synthesia is for presenter-led explainers rather than generative footage. Verdict: the most capable and most transparently documented choice here for filmmakers and content teams; the Free credits are enough to find out whether your shots work.

HeyGen

AI video with digital avatars

Create professional videos with AI avatars and voice cloning. Translate videos into 40+ languages with lip-sync. No camera or studio needed.

ElevenLabs

AI voice synthesis and voice cloning

AI voice synthesis. Clone any voice from seconds of audio, generate speech in 29 languages, and power podcasts, audiobooks, and content.

Canva AI

Magic Studio design tools for non-designers

Canva's AI is not a separate product; it is the Magic Studio set folded into the design tool a very large number of non-designers already use. Magic Design lays out a draft from a prompt or an upload, Magic Eraser and the background remover clean up photos, text-to-image produces illustrations, and the presentation and video generators turn an outline into slides or a short clip — all inside the same editor as the templates, brand kit and social scheduler, which is the whole argument for it. The plan structure is simple: a Free plan, Canva Pro for one person, which Canva's pricing page listed at US$144 a year when we read it on 30 September 2026, and a per-person Business plan for teams priced on the same page; Canva's pricing page blocked our automated read on 16 October 2026, so confirm the figures for your country before you buy. The honest limits: the AI tools carry usage quotas that differ by plan, generated images and video have the recognisable look of the category, and a team with a real brand system will want the Business tier for shared assets and approvals rather than the Pro plan. Midjourney, also on this site, is the choice when image quality matters more than speed, and Runway when the output is video. Verdict: the best budget option for social-media managers and small businesses who need good-enough visuals every day; not a replacement for a designer on anything that carries the brand.

Operator (OpenAI)

OpenAI's browser agent, now agent mode in ChatGPT

Operator no longer exists as a separate product, and that is the first thing anyone searching for it should know. OpenAI introduced it on 23 January 2025 as a research preview for Pro users in the US: an agent with its own browser, powered by a model OpenAI called Computer-Using Agent, that looks at a page through screenshots and types, clicks and scrolls to finish a task such as filling in a form or ordering groceries. An update dated 17 July 2025 on the same announcement page says Operator is now fully integrated into ChatGPT as ChatGPT agent, reached by choosing agent mode in the composer, and that the standalone operator.chatgpt.com site would be sunset; the page now points readers to OpenAI's workspace agents and ChatGPT Work instead. The design choices described at launch are still the useful part: the agent hands control back when it gets stuck, it is trained to ask you to take over for logins, payment details and CAPTCHAs, and you can take over the remote browser at any moment. The limits follow from that. It is slower than doing a familiar task yourself, it needs supervision at exactly the steps where a mistake costs money, and which ChatGPT plans include agent mode, with what usage, is set on OpenAI's own pricing page, which did not render for us on 16 October 2026 — so this page lists no price. Claude for Chrome on this site works inside your own browser rather than a remote one; Browse AI and Axiom AI are better for repeatable scraping and form-filling you want on a schedule. Verdict: treat this page as a pointer. Try ChatGPT's agent mode on your own one-off tasks, and use a dedicated automation tool when the job repeats.

Claude for Chrome

AI in every webpage

Anthropic's Chrome extension that brings Claude directly into any webpage. Summarise, write, and get help in context without switching tabs.

MultiOn

Early web agent whose site now redirects to AGI, Inc.

MultiOn was one of the early consumer web agents: you described a task such as a search, a booking or a purchase, and it navigated websites to finish it, with an API for developers building their own agents. What a buyer finds today is different. On 16 October 2026 multion.ai returned a permanent redirect to theagi.company, the site of AGI, Inc., which presents AGI-0 — an agent that sees a phone screen and acts across Android apps without APIs or integrations, offered through Google Play — and an AGI Platform for developers building action agents. That page does not mention MultiOn by name and shows no prices, so the $25-a-month plan this directory used to list cannot be confirmed and has been removed. The practical consequences: there is no MultiOn sign-up page, no published plan, and nothing on the vendor's own site about what happened to existing MultiOn API users. Readers who came for browser automation have clearer options here. Claude for Chrome works inside your own browser; ChatGPT's agent mode is what replaced OpenAI's Operator; and Browse AI and Axiom AI handle repeatable scraping and form-filling with prices on their own pages. Verdict: do not build a workflow on MultiOn today. If the AGI-0 Android agent interests you, judge it on its own page and terms, not on MultiOn's history.

Browse AI

No-code scraping and website monitoring, metered in credits

Browse AI turns a website into a spreadsheet without code. You point a robot at a page, show it which fields to capture, and it extracts them on a schedule, or watches the page and alerts you when something changes. The meter is the credit: the pricing page (read 16 October 2026) says each credit extracts 10 rows of data from a page or captures one screenshot, tasks on sites it classes as premium cost a minimum of 2 to 10 credits, and unused credits reset at the end of each billing cycle. The plans: Free with 50 credits a month, two websites and three users; Personal at $19 a month billed annually, with 12,000 credits a year and five websites, or $48 month to month for 2,000 credits a month; Professional at $69 a month billed annually, with 60,000 credits a year, ten websites and ten users, or $87 monthly for 5,000 credits a month; and Premium from $500 a month billed annually. Below Premium, extracted data is kept for 90 days, a single task can run for up to 60 minutes, and monitors check no more often than hourly on Free and every five minutes on paid plans. The limits: monthly and annual billing carry different credit allowances, so compare plans on rows per month rather than headline price; protected sites burn credits faster; and you remain responsible for respecting each site's terms. Axiom AI on this site is the better fit when the job is clicking through a process rather than collecting data, and ChatGPT's agent mode suits one-off tasks. Verdict: the simplest tool here for watching competitor prices or listings on a schedule. Start on Free, count the rows you really need, and size the plan from that.

No-code browser bots, priced by runtime

Axiom AI automates the clicking rather than the reading. You build a bot from steps — open a page, fill a field, click, loop through a spreadsheet, pull a table — and it repeats them in Chrome on your own computer or in Axiom's cloud. Pricing is by runtime rather than by task: the pricing page (read 16 October 2026) lists Starter at $15 a month for 5 hours, Pro at $50 for 30 hours, Pro Max at $150 for 100 hours and Ultimate at $250 for 250 hours, with only monthly prices shown. New accounts get 2 free hours of runtime without a card, and that trial allowance does not renew. Which plan you need depends on where the bot runs. Desktop runs have no time limit and unlimited concurrency on every plan, while cloud runs are capped per run — 1 hour on Starter, rising to 12 on Ultimate — and Starter has no cloud scheduling at all; Pro schedules cloud runs daily and Pro Max hourly. Webhooks, the API, Zapier and Make triggers and the desktop MCP server start at Pro, and team accounts are Ultimate only. The limits: Starter is in practice a desktop tool; any bot breaks when a site changes its layout; and the plan cards and feature table disagree on Ultimate's cloud concurrency and scheduling, so confirm before buying for scale. Browse AI on this site is the better buy when the goal is collecting data on a schedule, and ChatGPT's agent mode handles one-off tasks without building anything. Verdict: good value for operations staff and virtual assistants who repeat the same web process every day. Prove the bot on the 2 free hours, then choose between desktop runs on Starter and scheduled cloud runs on Pro.

Synthesia

Presenter-led AI video from a script

Synthesia solves a specific problem: a training or explainer video with a person on screen, without filming anyone. You write or paste a script, choose an AI avatar and a voice, and it renders a presenter delivering it, with dubbing into other languages. The pricing page (read 16 October 2026) now meters everything in credits. Basic is free with 500 credits a month — about 10 minutes of video creation or dubbing — nine avatars and a Synthesia watermark. Starter is $29 a month billed monthly or $18 billed yearly for 1,250 credits a month, up to roughly 12 more minutes, 125+ avatars, one editor and three guests. Pro is $89 monthly or $64 yearly for 6,000 credits, up to about 60 more minutes, 180+ avatars and five personal avatars. Enterprise is custom, with unlimited video minutes, SSO and SCORM export for learning-management systems. Every plan lists 140+ languages and voices. The limits: the minute figures are approximate conversions of credits, so long scripts cost more than the headline suggests; SCORM export, which most corporate training teams need, appears only on Enterprise; and avatars still read as synthetic to many viewers, which matters more in customer-facing marketing than in internal training. The page is also inconsistent about Starter's yearly credit total, so check it at checkout. HeyGen on this site is the closest rival for avatar video and translation, and Descript is for editing footage you filmed yourself. Verdict: the default here for HR, onboarding and internal-communications teams producing a steady flow of presenter videos — find out whether you need SCORM before you choose a plan.

Descript

Edit audio and video by editing the transcript

Descript's idea is that a recording is easier to edit as a document. It transcribes your audio or video, and deleting a sentence from the text cuts it from the recording; screen recording, multitrack editing and AI voice tools sit around that core. The pricing page (read 16 October 2026) lists five plans, priced per person per month: Free at $0 with 1 media hour, 100 one-time AI credits and 720p watermarked exports; Hobbyist at $16 or $24 with 10 media hours, 400 AI credits a month and 1080p exports without a watermark; Creator at $24 or $35 with 30 hours, 800 credits, 4K export and one to three seats; Business at $50 or $65 with 40 hours, 1,500 credits, more stock voices and translation proofreading; and Enterprise on a custom quote. The page shows each pair of prices under a monthly and annual toggle without saying which is which, but its sign-up buttons lead to annual checkout, so the lower figure appears to be the annual rate. Voice clones, text-to-speech and Regenerate are included from Hobbyist, and the page warns that past a plan's limit generated speech turns into 'jibber jabber'. The limits: media hours and AI credits are separate meters and both run out; the transcript-first approach is excellent for talking-head video and podcasts and less natural for effects-heavy edits; and the Free plan is for testing, not publishing. Synthesia on this site generates a presenter instead of editing one, and Runway generates footage from prompts. Verdict: the most efficient editor here for podcasters, course creators and marketers whose material is mostly speech — Hobbyist covers a weekly show, Creator a small team.

Glean

Permission-aware AI search and agents across work apps

Glean is enterprise search with an AI layer on top: it connects to the apps a company already uses and answers questions from all of them at once, while respecting who is allowed to see what. Its pricing page (read 16 October 2026) publishes no price, plan names, seat minimum or trial — the only call to action is 'Get a demo' — and says it connects to 275+ work apps and data sources, naming Slack, Google Drive, Jira, Confluence, SharePoint, GitHub and Salesforce. Access is described as permission-aware from day one, so people only see what they are already allowed to see. The same page cites customers building agents on it at scale — 3.4k+ at Zillow and 2.7k+ at Ericsson — which shows where it is aimed: large organisations whose knowledge is scattered across dozens of systems. In practice the value grows with how fragmented your tools are, and so does the risk: search makes every badly set sharing permission visible, so a permissions clean-up usually comes first. The limits: no self-serve route, no published price to budget from, and an implementation that needs IT and security sign-off. For a team that already keeps its knowledge in Notion, the workspace Q&A in Notion's Business plan covers much of the same need inside one tool; a Microsoft 365 organisation should compare Copilot for Microsoft 365 first. Verdict: a serious option for enterprises with sprawling knowledge and an IT team to run it; small and mid-sized companies should exhaust the search built into tools they already pay for.

n8n

Workflow automation you can run on its cloud or host yourself

n8n is the automation tool for people who would rather own the workflow engine than rent it. You build workflows visually from nodes, drop into code where a node is not enough, and run them on n8n's cloud or on your own server. The pricing page (read 16 October 2026) quotes euro prices billed annually: Starter at €20 a month for 2,500 workflow executions and 5 concurrent executions; Pro at €50 for 10,000 executions and up to 50 concurrent; Business at €667 for 40,000 executions, self-hosted only for now, with a 14-day trial that needs a card; and Enterprise by contact, with 200+ concurrent executions. The cloud plans offer a free trial without a card, and every plan lists unlimited active workflows and unlimited steps. The metering is the main reason to consider it: an execution is one run of the whole workflow, however many steps it has and however much data it moves, so long multi-step workflows cost far less than under per-step pricing. The Community edition is the standard self-hosted version on GitHub, released under n8n's fair-code licence rather than a conventional open-source one, so read the licence before you embed or resell it; paid self-hosted licence keys must ping n8n's licence server daily. The limits: self-hosting means you own uptime, upgrades, backups and security; Starter's 2,500 executions go quickly on a workflow triggered every few minutes; and prices are in euros. Zapier on this site is the simpler, maintenance-free choice; Make sits between the two. Verdict: the best value here for technical teams running many multi-step workflows — self-host if someone can maintain it, otherwise start on cloud Pro.

Notion AI: Workspace Q&A

Ask questions across your workspace and connected apps

This page covers one slice of Notion AI: asking a question in plain English and getting an answer drawn from your workspace and, increasingly, from the apps connected to it. It is not a separate product, and the $10-a-month add-on it was once sold as no longer appears on Notion's pricing page (read 16 October 2026). There, Free ($0) and Plus ($10 per member a month) carry only a limited trial of the AI features — Notion Agent, meeting notes, Enterprise Search and Research mode — while Business, at $20 per member a month, includes them, and Enterprise is custom-priced with zero data retention at the AI model providers. For search specifically, Business lists searching across connected apps such as Slack and GitHub as a beta, and Notion marks Jira, Box, OneDrive, Salesforce and Asana as beta connectors; Research mode, also in beta, uses deep reasoning to produce longer reports. The page lists no caps on queries or documents. The practical test is simple: answers are only as good as what is written down and connected, so a workspace of stale pages produces confident, stale answers. The limits: connected-app search is still beta, the full feature needs members on Business, and annual billing saves 'up to 20%' without per-member annual prices shown. The main Notion AI card on this site covers writing, agents and autofill; Glean is the enterprise-wide alternative when knowledge lives in many systems; Mem suits individuals who would rather not file at all. Verdict: worth it for a team whose wiki and projects already live in Notion and are kept current; not on its own a reason to move to Notion.

HubSpot AI

CRM with AI built throughout

HubSpot's AI features span the entire platform — AI email writing, content generation, chatbots, predictive lead scoring, and conversation intelligence.

Fireflies.ai

Meeting recorder, transcripts and AI notes for every call

Fireflies joins your Zoom, Google Meet or Microsoft Teams calls, transcribes them and turns each one into a searchable record with a summary, action items and an AI assistant, AskFred, you can question afterwards; a Chrome extension can also record without a bot joining. The pricing page (read 16 October 2026) lists Free at $0 for ever, with transcription, limited AI summaries, 400 minutes of storage for the whole team and 20 AI credits; Pro at $10 per seat a month billed annually or $18 monthly, with unlimited summaries, 8,000 minutes of storage per seat, video recording, downloads and integrations; Business at $19 annually or $29 monthly, with unlimited storage, 30 AI credits, conversation intelligence, team analytics and multi-language mode; and Enterprise at $39 per seat, annual only, adding SSO and SCIM, HIPAA compliance, custom data retention and private storage. Transcription is listed in 100+ languages. Each recording is capped in length — two hours on Free and Pro, three on Business, four on Enterprise — and AI credits are a separate allowance on every plan. The limits: Free storage is shared by the team and fills quickly; a bot in the call can put off external guests; and the summaries are good starting notes, not minutes to sign off unread. Otter.ai on this site is the closest rival for meeting transcription, and Reclaim AI handles the calendar around the meetings. Verdict: with a free plan and Pro at $10 per seat a month billed annually, an inexpensive way to give a sales or consulting team searchable call records — test accuracy on your accents and jargon on Free, then move to Pro for storage and integrations.

Writer

Enterprise AI grounded in company knowledge and brand voice

Writer is built for the company rather than the individual writer. It grounds generation in your own material through a Knowledge Graph, applies house voice through personality profiles and style guides, and increasingly sells agents: its plans page (read 16 October 2026) describes 'WRITER Agent — one interface for chat and automation' on Starter and an Agent Library of 100+ prebuilt agents on Enterprise. There are two plans and no published prices. Starter is for up to five users on monthly or annual per-seat plans with fixed credit limits, with a 14-day free trial that needs no card, one team personality profile, limited access to AI Studio, WRITER's default models and 50 GB in a single Knowledge Graph. Enterprise adds unlimited users and free collaborators, departmental brand and voice profiles, full AI Studio, full Knowledge Graph capabilities with unrestricted connectors, and the option to work with your preferred large language models. Because the per-seat price is not on the page, this directory no longer shows the $18-a-month figure it used to list; ask for the Starter quote during the trial. The limits: the value depends on the time you invest in the Knowledge Graph and voice profiles; Starter's five-user cap pushes any real team into an Enterprise conversation; and it is far more than a solo writer needs. Copy.ai on this site is the go-to-market alternative with a published self-serve price, while Wordtune and Sudowrite serve individual writers. Verdict: a strong fit for marketing and content teams in regulated or brand-sensitive companies with the resources to set it up; overkill for small teams.

SEO.ai

Done-for-you AI SEO agent, priced per website

SEO.ai has turned from an article writer into what its pricing page (read 16 October 2026) calls an 'AI agent that works 24/7' for organic growth. It recommends content and a publishing plan, writes pages and articles, says it targets visibility in both Google and ChatGPT, offers backlinks 'from real non-spammy websites', builds and optimises Google Ads, monitors performance and includes human support. Two plans are listed: Single Site at $149 a month for one website in one language, and Multi Site at $299 a month for up to three websites or languages, with larger selections available. Annual payment saves 25%, prices exclude VAT, and there is a 7-day trial. The page states no article or credit allowance and no user count, so ask how much it will publish each month before you commit. The judgement call is the backlinks. Google's spam policies define link spam as links created mainly to manipulate rankings, list buying or selling links for ranking purposes as the first example, and accept paid links only when they carry rel="sponsored" or rel="nofollow" — so ask exactly how the links are obtained and how they are marked before switching that service on, because a penalty costs far more than the subscription. The other limits: AI-written pages need an editor who knows the subject, and an automated publishing plan can fill a site with thin pages. Surfer SEO on this site is the tool for scoring drafts you write yourself, and Writesonic for tracking how AI answers mention your brand. Verdict: a done-for-you package for a small business with no marketing staff; anyone protective of their domain should keep a human approving every page and every link.

Humata

Ask questions about your own documents, priced by pages

Humata is for reading documents you already have rather than searching the web. You upload PDFs and other files, ask questions, and it answers from the text, with summaries and comparisons across files. Its pricing page (read 16 October 2026) meters by pages processed: Free at $0 for one user and 60 pages a month, without OCR; Expert at $9.99 a month for up to three people and 500 pages, then $0.02 a page; Team at $49 per user a month for up to ten people and 5,000 pages, then $0.01 a page, adding OCR for scanned documents, department and folder permissions and a SOC 2 certificate; and Enterprise on custom terms with an uptime SLA. The FAQ also mentions a self-serve Student plan without showing its price, and the page lists GPT-5 support on the free tier. What the page metering means in practice: one long contract or annual report can use most of a month's Free allowance, and scanned PDFs need Team for OCR. The limits: answers are only as reliable as the text extraction, so check any figure against the page it came from; no annual pricing is shown; and confidential files call for the Team plan's controls. Elicit on this site is the better tool for academic papers you have not yet found, and Perplexity for questions about the open web. Verdict: an inexpensive document assistant for professionals who live in PDFs — Expert suits a small practice, and Team is worth it only when you need OCR and permissions.

HeyReach

LinkedIn outreach across many sender accounts, priced per sender

HeyReach automates LinkedIn outreach across many sender accounts from one place, which is why agencies and outbound teams choose it over single-profile tools. You connect LinkedIn accounts as senders, build campaigns of connection requests and messages, and handle every reply from a unified inbox without switching accounts; native integrations with Instantly and Smartlead add email, and an API and webhooks are available. Pricing is per sender, not per user — teammates are free. The pricing page (read 16 October 2026) lists Growth at $79 per sender a month, or $63 billed yearly, with 100 enrichment credits per sender; Agency at $999 a month for 25 senders ($799 billed yearly) with 1,000 credits and white-labelling for one brand; Unlimited at $2,999 a month ($2,399 yearly), capped at 300 senders under a fair-use policy, with 3,000 credits; and a custom done-for-you service. Quarterly billing takes 10% off. The FAQ offers a 14-day trial with three LinkedIn accounts and no card, and a 30-day trial on request for the agency plans, which expect you to bring your own proxies. The account risk sits with you: the page promises that each sender stays within daily limits and safe usage levels but does not publish those limits, and LinkedIn's own terms decide what automation it tolerates, so read them before connecting a valuable profile. Clay on this site builds and enriches the lead list that feeds a campaign; Amplemarket bundles data, sequences and email in one platform. Verdict: the efficient choice for agencies running LinkedIn outreach for many clients; a founder with one profile is better served by careful manual outreach.

Seamless.AI

B2B contact data, intent signals and a dialer, priced by quote

Seamless.AI sells B2B contact data: search for people or companies and it returns emails, direct dials and mobile numbers, with a Chrome extension that finds the same details while you browse LinkedIn or any other site. Around that core, its site (read 16 October 2026) lists buyer-intent signals, job-change tracking, CRM enrichment, AI-written pitches and AI fields, built-in emailing, an AI dialer, social selling, an API, an MCP server, apps inside Claude and ChatGPT, and integrations it puts at 6,000+ tools. What the pricing page does not show is a price. It is a 'Pricing Inquiry' form headed 'Perfect Pricing for Startups to the Fortune 500', offering special pricing for unlimited users, bulk credits, intent, job changes and enrichment, while a 'Sign up free' button sits in the site navigation; the free allowance is not stated on the pricing page. So expect to start free and to need a sales conversation for anything a team would use daily. The limits worth weighing: contact data is metered in credits, accuracy is something only a test on your own target accounts can establish, and you remain responsible for following the privacy and anti-spam rules that apply where your contacts live. Amplemarket on this site publishes a starting price for a comparable data-plus-sequences bundle, and Clay draws on many data providers rather than a single database. Verdict: worth the free sign-up to test its data on accounts you know; do not sign a contract until you have measured how many of the emails and numbers are actually right.

Agentic AI for enterprise marketing campaigns

Typeface has moved from a brand-safe content generator to what its homepage (read 16 October 2026) calls 'Agentic AI for enterprise marketing teams': software that orchestrates people, agents and systems to run on-brand campaigns at scale. Its components are named Arc Graph — brand guidelines, audiences, assets and campaign learnings held as structured intelligence that agents can act on — plus Arc Fabric, Arc Agents and Arc Loop. It connects through MCP, APIs and integrations, and works inside Figma, Photoshop and Illustrator. There is no pricing page (typeface.ai/pricing returned 404) and no free trial; the only calls to action are 'Get a demo' and 'Take a tour'. That tells you the intended buyer: a marketing organisation large enough to have brand governance, many campaigns in flight and designers already working in Figma and Adobe tools. The limits: no price to budget from; value that depends on loading your brand rules, audiences and assets into Arc Graph before anything useful comes out; and agent-made campaign material that still needs brand and legal review. Persado on this site is the specialist for scored, compliance-checked language in regulated sectors, Writer is the text-first enterprise alternative, and Smartly automates the media side once assets exist. Verdict: shortlist it if you run enterprise-scale campaigns and want agents working from one governed brand source; smaller teams will get further with Canva AI and a clear brand kit.

Pika

Text- and image-to-video generator, metered in credits

Pika generates short videos from text prompts and images and edits existing clips with AI. Its pricing page (read 16 October 2026) lists four tiers, all with access to every model and no watermark: Free at $0, which includes no monthly credits — you buy credit packs instead; Starter at $10 a month, or $8 billed annually, with 900 credits a month but no commercial licence; Creator at $35, or $28 billed annually, with 3,150 credits and a commercial licence; and Fancy from $95 a month ($76 annually), rising with the credit amount to $880, with early access to new models from the $440 tier. Annual billing takes 20% off, and the plan cards and comparison table differ slightly on some figures, so check at checkout. The licence line is the one to read: only Creator and Fancy include commercial use, so client work and monetised content need at least Creator. The limits: every generation and retry spends credits, clips are short, and the Free plan is a way to buy credits rather than a free allowance. Runway on this site is the established professional option with a free credit grant, PixVerse is another social-first generator, and Synthesia makes presenter videos instead. Verdict: a reasonable mid-priced generator for social clips — choose Creator if anything you make is for business.

Suno

Songs from a text prompt, with commercial rights on paid plans

Suno generates complete songs from a text prompt and can also work from audio you upload. Its pricing page (read 16 October 2026) lists Free at $0 with 50 credits a day, access to the free model v6-mini, no monthly song downloads, no commercial rights and audio uploads of up to eight minutes; Pro at $8 a month with 2,500 credits a month, 20 song downloads, the advanced v6 and v6-wild models, commercial use rights and two stem-separation types; and Premier at $24 a month with 10,000 credits, 60 downloads, the same models, commercial rights and three stem-separation types. Billing yearly saves $24 on Pro and $72 on Premier, according to the page, which does not print the annual totals. The rights line is the one to read twice: the Free plan carries no commercial rights, so anything you might publish, monetise or use in a client project should be made on a paid plan from the start. The limits: every generation and regeneration spends credits; a finished track still needs a listener's judgement on structure and mix; and on the paid plans the monthly download cap, not the credits, may be what stops a prolific user. ElevenLabs on this site handles voice and speech rather than music, while Runway and Descript cover the video a track might sit under. Verdict: the quickest way here to a usable song or jingle — prototype on Free, then switch to Pro before you make anything you intend to release.

Enterprise AI coding platform, now owned by Tricentis

Tabnine is an AI coding platform that its new owner describes as purpose-built for secure, context-aware enterprise software development — and its ownership is the first thing a buyer needs to know. On 30 July 2026 Tricentis, a software-testing company, announced that it had acquired Tabnine and would integrate Tabnine's Enterprise Context Engine into the Tricentis Agentic Quality Engineering Platform to strengthen its testing agents. The release describes that engine as a structured, continuously updated knowledge graph of an organisation's systems, built from repositories, documentation, tickets, APIs and infrastructure, and lists deployment on-premises, in a private VPC or fully air-gapped. On 16 October 2026 both tabnine.com and its pricing page redirected to Tricentis pages, so there is no Tabnine plan or price to read, and the $12-a-month figure this directory used to list has been removed. What that means in practice: existing customers should ask Tricentis directly about their contract and roadmap, and new buyers should treat Tabnine as part of a testing platform rather than a standalone editor plug-in. GitHub Copilot on this site is the mainstream in-editor assistant with published prices, Claude Code and Cursor are the agentic choices, and Cohere suits teams that need models deployed privately. Verdict: not a product to start with in 2026 unless you are already buying from Tricentis; compare the Copilot and Cursor cards instead.

Microsoft 365 Copilot

AI add-on licence for the Microsoft 365 apps

Microsoft's Copilot for work is an add-on licence that puts AI inside the Microsoft 365 apps an organisation already pays for, such as Word, Excel, PowerPoint, Outlook and Teams, and grounds work answers in what each user's account can already access. Microsoft's licensing documentation (updated 30 September 2026) sets out the structure. Copilot is bought as an add-on to an eligible plan: Microsoft Copilot Business goes with Microsoft 365 Business Basic, Standard and Premium, Microsoft 365 Apps for Business and Teams Essentials, while the enterprise Copilot licence can be added to a long list that includes Microsoft 365 E3 and E5, Office 365 E1, E3 and E5, the frontline F1 and F3 plans, and standalone Exchange, SharePoint, OneDrive, Teams, Planner, Project and Visio plans; Microsoft 365 E7 includes Copilot. Education and US government versions exist. Separately, Copilot Chat with web-grounded answers is included at no extra cost with an eligible Microsoft 365 subscription, while work-grounded chat needs the Copilot licence. The same page notes that the Microsoft 365 Copilot app is now called Microsoft Copilot. Microsoft's pricing pages did not load for us on 16 October 2026, so the $30-a-user figure this directory used to show has been removed; get the current price for your plan and country from Microsoft or your reseller. The limits: Copilot can surface anything a user can technically open, so loose sharing permissions become visible; its value depends on how much work really happens in Microsoft 365; and the free web-grounded chat may already cover casual use. Glean on this site is the cross-application search alternative, and Notion AI the equivalent for Notion-first teams. Verdict: the default choice for organisations standardised on Microsoft 365 — tidy permissions, pilot with a small group, and compare against free Copilot Chat before licensing everyone.

Wordtune

Sentence rewrites, grammar checks and summaries

Wordtune is a rewriting tool before it is a writing tool: you highlight a sentence and it offers alternative phrasings, alongside grammar and spelling checks, summaries of long articles and documents, AI generation and a 'Humanize AI' feature. Its plans page (read 16 October 2026) lists Basic at $0 with 10 rewrites and AI suggestions a day, three AI summaries a month and unlimited spelling and grammar checks; Advanced at $6.99 a month, or $4.89 a month billed annually, with 30 rewrites a day and 15 summaries a month; and Unlimited at $9.99 a month, or $6.99 billed annually, with unlimited rewrites and summaries plus vocabulary, clarity and fluency suggestions and premium support. Both paid plans start with a 3-day free trial. Chrome and Edge extensions and an iOS app are linked from the site, though the page does not say which plans include them, and business features such as SAML SSO and centralised billing are marked unavailable on all three plans. The limits: the daily rewrite cap is what Basic and Advanced users hit first; rewording a sentence does not repair a weak argument; and a 'humanizer' does not make AI-written text your own work, so check what your school, employer or client allows before using it. Sudowrite on this site is built for fiction, Writer for enterprise teams and Copy.ai for go-to-market content. Verdict: an inexpensive daily helper for anyone polishing emails, reports and essays in English — Unlimited on annual billing is the plan to choose if you rewrite every day.

Speechify

Text-to-speech reader for documents, web pages and scans

Speechify reads text aloud — web pages, documents, PDFs and photographed pages — in apps for iPhone and iPad, Android, Chrome, Edge, the web, Mac and Windows. People use it to get through a reading load faster, to rest their eyes, or because reading on a screen is slow or tiring for them. Its pricing page (read 16 October 2026) lists a Free plan with 10 robotic-sounding voices and listening speeds up to 1.5x, and Premium at $29 a month on monthly billing, with 1,000+ natural-sounding voices in 60+ languages, speeds up to 5x, Scan & Listen for printed pages, AI summaries and chat, Google Drive, Dropbox and OneDrive integrations, voice typing, AI podcasts and a voice assistant. Yearly billing is labelled 'save 60%', but the page does not print the yearly price, so the $139-a-year figure this directory used to list has been removed; check the annual price at checkout. Studio, for voice-over production, and the API are priced on separate pages. The limits: the jump from Free to Premium is large because the natural voices are the product; very high speeds take practice to follow; and it reads rather than explains, with the AI summaries as the exception. ElevenLabs on this site is the tool for producing voice-overs rather than listening, and Otter.ai goes the other way, from speech to text. Verdict: genuinely useful for heavy readers, commuters and anyone who absorbs text better by ear — try the free voices first, and buy annually if it sticks.

Loom AI

AI-enhanced screen recording

Record and share videos with AI-generated titles, summaries, and chapters. Async video communication that replaces meetings.

Cohere

Enterprise models and platform, with private deployment

Cohere sells AI to enterprises rather than to consumers, and it competes on where the model runs as much as on the model itself. Its pricing page (read 16 October 2026) lists four products: North, described as an all-in-one AI platform for automating tasks safely across an enterprise; Command, its generative language models; Embed, a semantic representation model for search; and Rerank, for retrieval optimisation. API calls made from a trial key are free, but trial keys are rate-limited and not permitted for production or commercial use. The per-token prices printed on the page are labelled as legacy models for existing customers — Command at $1.00 per million input tokens and $2.00 per million output, Command R+ 08-2024 at $2.50 and $10.00, Command-light at $0.30 and $0.60, Aya Expanse at $0.50 and $1.50 — and Rerank is billed per search unit, one query with up to 100 documents, without a figure shown. Current models and enterprise work are priced on request, including private deployment on-premises or in an isolated VPC and bespoke model customisation. That positions it clearly: for regulated organisations that want retrieval-augmented search and generation without sending data to a shared cloud. The limits: no published price for the current line-up, legacy rates that may not apply to a new contract, and a developer-first product that needs engineering time before anyone else benefits. On this site, Claude Code and GitHub Copilot are coding assistants rather than model platforms, and Glean is the packaged enterprise-search option for teams that would rather buy than build. Verdict: a strong choice for enterprises building search or assistants on private infrastructure — start with a trial key, and get a quote before you set a budget.

Synthflow

AI phone agents, sold on enterprise contracts

Synthflow builds AI voice agents that answer and place phone calls — inbound support, outbound qualification, appointment booking — and it has moved firmly up-market. Its pricing page (read 16 October 2026) lists a single Enterprise plan: contracts start at $30,000 a year, with the final price set by call volume, concurrency, telephony, integrations, security needs and launch support. Telephony can run on Synthflow's own numbers, your SIP trunk or approved enterprise telephony; concurrency is planned per contract; and the page shows SOC 2, HIPAA, GDPR and ISO 27001 badges, EU and US hosting, and support for MSA and DPA paperwork and enterprise security reviews. It lists no per-minute rate, no included minutes and no free trial, so the $29-a-month self-serve plan this directory used to show has been removed. What that means: Synthflow is now a contract purchase for contact centres and businesses with steady call volume, not something a small business can switch on for an afternoon. The limits: the $30,000 floor; voice quality and response speed that you can only judge on a pilot with your own callers; and the consent and recording rules that apply to automated calls wherever your callers are. Intercom Fin and Zendesk AI on this site handle chat and email first, and Tidio suits small businesses whose customers write rather than call. Verdict: shortlist it for high-volume phone operations that need compliance paperwork and a choice of hosting region; everyone else should start with a chat agent.

Flexible CRM with AI agents, metered in credits

Attio is a CRM you shape like a database: companies, people and deals are objects you can extend with your own, and AI — Ask Attio and AI agents — works across the data, metered in credits that are shared with workflows. The pricing page (read 16 October 2026) lists Free at $0 for up to three seats, with 50,000 records, three objects, 200 emails a month, 100 credits per seat and 250 workspace credits; Plus at $44 per seat a month, or $35 billed annually, for up to ten seats, 250,000 records, five objects and 1,000 emails a month; Pro at $99, or $79 billed annually, with unlimited seats, a million records, twelve objects, unlimited email and 10,000 workspace credits; and Enterprise on custom annual terms. Extra workspace credits come in packs, starting at 5,000 for $85 a month. Workflows, sequences and call intelligence appear in the comparison table on every plan, but the Pro card presents call intelligence and sequences as Pro features, so confirm which plan you need before buying for them. The limits: credits are shared between AI and automation, so a busy workflow can starve the AI and the reverse; the flexible data model takes design time up front; and Free's three-seat cap and email allowance make it a trial for most teams. HubSpot AI on this site is the all-in-one marketing-and-sales alternative, Clay enriches lists before they reach a CRM, and Amplemarket bundles data with sequences. Verdict: a strong modern CRM for start-ups, investors and agencies willing to model their own pipeline — begin on Free and move to Plus when the team grows past three.

Social media scheduler your AI agent can post through

Postiz is a social media scheduler built for AI agents as much as for people. Its pricing page (read 30 October 2026) says every plan includes a hosted MCP server, a command-line tool, the public API and webhooks, so Claude, ChatGPT, Cursor or any other agent that speaks MCP can draft and schedule posts for you; there is also a built-in Smart Agent you instruct by chatting. It covers the major networks, among them X, LinkedIn, Instagram, Facebook, TikTok, YouTube, Threads, Pinterest, Reddit, Discord, Bluesky and Mastodon. Standard is $29 a month for five channels, Team $39 for ten channels with unlimited team members, Pro $49 for 30 channels and Ultimate $99 for 100; every plan has unlimited posts, and the allowance of AI images, AI videos and video-clipping minutes rises with the plan. There is no free plan, only a seven-day free trial, and the code is open source so it can be self-hosted, although Postiz says the hosted platform is where agent features ship first. The limits: Standard has no team support, refunds cover at most one month of fees and annual plans are not refundable, and an agent that publishes on your behalf still needs someone checking what goes out. Verdict: the scheduler to pick if you want an AI agent, not a person, doing the posting; if you only need a calendar for a few posts a week, a plain scheduler will do.

Squad

AI teammates on your own AI plan, $99 a month flat

Squad (formerly MissionControlHQ) sells a team of AI agents rather than a single assistant. Its pricing page (read 30 October 2026) lists one plan at $99 a month: an onboarding in which an AI Chief of Staff learns your business and designs the squad with you, unlimited AI teammates you name and give roles, a dedicated cloud computer with a separate browser profile for each teammate, more than 70 integrations with per-teammate access controls, a secret vault for keys and passwords, a ticket queue with one-tap approvals and daily caps, and a full log of every run. The work runs on an AI plan you bring, such as ChatGPT, Claude and Claude Code, Gemini, SuperGrok or Copilot, and Squad recommends a ChatGPT plan at $100 to $200 a month on top of its own fee. Extras are optional: an email inbox per AI teammate at $10 a month, seats for human colleagues at $9, research credits beyond the 1,000 included each month, and a second squad at another $99. There is no free tier, but there is a seven-day money-back guarantee. The limits: the real monthly cost is two subscriptions, not one; an agent with access to your tools needs its scopes and approvals set before it runs; and it pays off on repeatable work, not on occasional questions. Verdict: worth trying for a founder or operator who already pays for a top-tier AI plan and wants it working through a queue of jobs; without one, Claude or ChatGPT on its own is the cheaper first step.