TODO for AI
vs 
Claude Cowork
By Marcell Havlik, founder of TODO for AI · Updated
Anthropic's agent, merged into Claude itself in September 2026: ask for anything and Claude decides whether it is an answer or a task it works through on its own. Here is where each one wins, with numbers, and who should pick which.
| Fact | Claude Cowork | TODO for AI |
|---|---|---|
| Interface | Claude Desktop, web, mobile, Chrome side panel | Web, desktop, mobile, CLI, IDE (Zed, JetBrains, VS Code), voice |
| Models | Anthropic only | Any provider: Claude, GPT, Gemini, Grok, Ollama |
| Source | Closed | Edge + CLI public |
| Starts at | $20/mo | Free |
Key takeaways
- Models. Claude Cowork model choice: Anthropic only. TODO for AI runs any provider, Claude, GPT, Gemini, Grok or a local Ollama model, on one plan, and one model can review another's work.
- Benchmark. TODO for AI scored 89.9% on Terminal-Bench 2.1 with Opus 5.5 (high) (own harness, 80/89 tasks, 2026-10-08; not an official leaderboard submission).
- Price. Claude Cowork pricing: Included on every paid Claude plan: Pro $20, Max $100 or $200, Team, Enterprise. TODO for AI has a free plan (Claude Sonnet only), then Starter $20, Pro $100 or Ultra $200 a month with any provider included.
- Where the work happens. Claude Cowork gets one task done in Anthropic's sandbox. TODO for AI is one board your team and agents work through together, on your own PC, a cloud VM and your logged-in Chrome.
- Who should pick which. Pick Claude Cowork if Anthropic models are all you need. Pick TODO for AI if you want Claude, GPT and Gemini on one task, reviewing each other before anything ships.
Claude Cowork
Anthropic only
TODO for AI
Any providerWe are not competing for the same job
Claude Cowork gets the task done.
TODO for AI organises the whole company’s work.
Claude Cowork answers one question: how do I get this task done, in Anthropic's sandbox? So do the others — Cursor inside the editor, Claude Code in one person's session, Lindy by asking its assistant.
We answer a different one: where does the work live? One board per project, and a project for each part of the company — engineering, marketing, ops, finance, one per client. Your colleagues work on it. So do your agents, as many as you want to run, each with its own model, tools and permissions. Items carry the people who worked them and the agent that ran them, and the board filters by person, so what the company has in flight is one screen instead of a Slack thread.
That is not a team plan bolted onto an AI tool. It is the product. The agents are how items get closed; the board is what the company runs on.
What Claude Cowork is, and where it stops
Cowork is Anthropic taking Claude Code's autonomy out of the repo and pointing it at everything else. Since 16 September 2026 it is no longer a separate tab: chat and Cowork are one Claude, which decides whether a request is a quick answer or a task. You hand over a multi-step task and it works through it on its own: organising files, producing documents, researching, clicking through sites with its built-in browser. Scheduled tasks run on Anthropic's servers, so work continues with your laptop shut, and it is included on every paid plan rather than sold as an add-on. It is the most capable autonomous agent any big model vendor ships today.
It is also, by design, a Claude product end to end, and a personal one: Team plans can share a Cowork project with view or edit access, but there is no board where a colleague and an agent hold the same task. Every step runs on Anthropic models, in Anthropic's sandbox, extended by MCP connectors. Which means no second vendor ever reviews the output, the agent never gets a shell on the machine your work actually lives on, and anything you want it to use has to exist as a connector rather than as the CLI tool you already have installed.
Claude subscribers who want autonomous document, research and browser work without leaving Anthropic.
Three reasons Claude Cowork users look elsewhere
Plan, execution and review all come from Claude. There is no way to have GPT or Gemini check the work before it ships, which is exactly what you want on the long autonomous runs Cowork is built for.
Sessions run in Anthropic's isolated environment: not a shell on your dev box, not your repo. Claude in Chrome does reach your logged-in browser from the side panel, but the rest of the run is still theirs, not yours.
Reach is its browser plus whatever MCP exposes. The CLI you already use for deploys, ad accounts or your CRM is not a connector, so Cowork cannot simply pick it up.
Six dimensions that decide it
Claude Cowork 14/30TODO for AI 29/30How does Claude Cowork compare to TODO for AI, feature by feature?
| Feature | Claude Cowork | TODO for AI |
|---|---|---|
Persistent cloud VM An always-on machine. Files, logins and sessions survive between runs. | Their sandbox | Yes |
Model choice | Anthropic only | Any provider |
Cross-vendor review Review GPT output with Claude, or the other way around. | No | Yes |
Drives your real browser Browser extension (Chrome, Firefox) plus cloud browsers. | Yes | Yes |
Business tasks, not only code SEO, email, ads, social, CRM via CLI tools. | Docs, research, browser | Yes |
Any CLI as an integration | MCP connectors | Any binary |
Voice control | No | Yes |
People and agents on one board Colleagues and agents work the same list. Items show who contributed and which agent ran them; filter the board by person. | No | Yes |
Many agents, each configured Each agent gets its own model, prompt, tools and permissions. Pick which one takes an item. | One assistant, skills | Unlimited, per task |
Projects for the whole company Engineering, marketing, ops, finance: separate projects and agents, one team and one bill. | Projects, shareable on Team | Unlimited, shared |
Team roles, pooled usage | Per seat | Roles + pooled |
Tool permissions | Permission modes | Allow, approve, block |
Open runtime | No | Edge + CLI public |
Same ladder, different meter
Rolling 5-hour windows plus a weekly cap, shared with Claude chat. An agentic run consumes many chat turns worth, so windows go fast. Paid plans can continue at API rates.
One pool priced in dollars of model cost, spendable on any vendor. Session and weekly windows pace it, and a 1.3-1.5x plan multiplier applies, so a $20 plan is about $15 of list-price API. Team seat prices go into one shared pool.