A research worker for Claude and ChatGPT. Your AI keeps the reasoning. Legwork's worker searches, reads and cross-checks your documents, then hands back a short report with sources.
Your AI doesn't get a pile of raw files. It gets a few paragraphs with a fixed shape:
One fact per line, each with the file and location it came from.
Short quotes, so the claim can be checked without opening the file.
What the worker looked for and couldn't find. Gaps are stated, not filled in.
Where does Mara first appear, and who introduces her?
"…my sister waved her over from the harbour wall and said, this is Mara."
No mention in part-2 drafts before chapter 9.
Ask about your manuscript like you'd ask an assistant who has read every chapter. Get the answer with the chapter and line. Our benchmark ran on a real 2,275-file manuscript.
Find what you already wrote, what you already promised, and what the brand guide says, with sources, before you write the next piece.
Cross-check dozens of documents. Every finding comes with the exact place it came from, and what wasn't found is listed too.
A remote MCP backend. Bring your own key and pick your worker model.
Shared workspaces and skills, with one budget per team.
The model you trust makes the judgment calls.
The exact file and location for every claim, so you can check it in seconds.
Any low-cost worker model you pick, or a local one. You pay the provider directly; we never resell tokens.
The same files and history in Claude and ChatGPT.
Built for people who don't want to configure agents.
Word, PDF, spreadsheets, slides, notes. No conversion needed.
Connect your own account with a low-cost model provider, or run a local model. You pay the provider directly, and you see which worker model read your documents. How your data is handled →
Delegated answers take about as long. The saving is in the work your main model does, not in waiting time.
That's why every finding carries a source, and why re-running the worker is cheap.
We'll publish results from inside the Claude and ChatGPT apps when that testing is done.
Tell us what you work on and which AI app you use. We'll email you when your spot opens.