Docs plus email plus private ChatGPT.
Several copies, comments detached from versions, fluent rewrites that add rules the sources do not support, and “looks good” in chat standing in for approval.
For the people who sign
Approving documents has become a lottery.
An AI agent drafted it. A colleague rewrote it in a private chat. Legal commented on yesterday's version. Now it is on your desk for a signature, and nobody can say who wrote what or which version this is.
8Hats Lab is building a shared room where people and their AI agents take one document from brief to an approved version, every contribution visible, a named person deciding. We are looking for two design partners to build it around your documents.
Apply as a design partnerPrefer a call? Book 20 minutesFill in the contact form
Two companies. Five fields. Tell us which document you would bring.
This is your situation if
The draft looks finished. The record does not exist.
A polished draft arrives and you cannot tell what a person wrote, what an AI agent drafted, and what anyone checked.
Comments live in email and chat. A late edit changed the meaning after the review, and silence got counted as approval.
You approved a version that should not have been approved, and it is your signature on it.
40%
of U.S. desk workers surveyed encountered AI-generated work that shifted interpretation and clean-up to a colleague in a single month. Each incident took nearly two hours to resolve.
Axios · BetterUp Labs + Stanford Social Media Lab · 1,150 U.S. desk workers · 2025AI is not the problem
The speed is real. The record is missing.
Try building a strategy paper from the numbers of five departments without AI agents: a week of evenings, and it is out of date when it lands. With them it takes an afternoon, and it is better. Nobody is going back.
What the speed removed is the slow, visible trail people used to leave by accident: who drafted, who objected, who signed off. Fast work needs that trail on purpose. A clear process, and a system that keeps the record.
What changes
What you can stand behind.
The document is not the problem. The missing record is. Four things change when people and AI agents work on it in one room.
One current version, with an ID. Copies stop. Every comment points at that version.
Every objection kept, with its owner. A disagreement stays beside the exact sentence, with its source, until a person closes it.
Silence shown as missing, not as approval. A required review that has not happened is visible as not happened.
One frozen version, approved by a named person. The record says who proposed, who changed, who decided.
- A person: Brief and owner
- An AI agent: Draft from sources
- An AI agent: Objections kept
- An AI agent: Missing review shown
- A named person decides: One version approved
Why the usual way keeps failing
Three tools, and none of them keeps the decision.
Document systems and e-signature.
They version the file. They are blind to who contributed, what was disputed and what was never reviewed.
An AI policy.
Rules without a working process. Nobody reads a policy at the moment they forward an AI summary.
Design partners, not a beta list
We want two companies to build this with us.
Two design partners will build this case with us around their documents, reviewers, and baseline. They shape the product and get terms nobody later gets.
What you get
You shape the product. A weekly working session with our team decides what gets built next, from your document’s real friction.
Your first cycle, configured with us. One real document, the shared room, the record, and the two company AI agents set up with your document owner.
90% off the first year. When the product launches, design partners pay a tenth of the first-year price.
The record and the templates stay yours, whatever you decide afterwards.
What we ask
One recurring document with a named owner and the people who must review it.
One working session a week with the document owner during the first cycle.
Your own baseline. Review cycles per document, reviewer hours, versions approved that should not have been: measured on your documents in week one, so the result is yours, not our average.
Honest feedback, including the parts that do not work.
Who fits: companies of roughly 50 to 2,000 people where consequential documents already arrive AI-drafted, and one leader in HR, People or operations who wants this fixed for a document they own.
What happens after you apply
You send the application
Who you are, which document, who owns and reviews it.
We schedule a call
We show what we do, on a document like yours, and discuss your task.
We decide together
Whether the partnership fits, and on what scope.
Either way, you get our working notes on how to change the document workflow for people and AI agents: the guide, the templates, and what we have learned running it ourselves.
Apply as a design partner
Apply as a design partner.
Tell us who you are and which recurring document you would bring. We read every application and follow up about the fit.
What if
The questions we get first.
“You will read my people’s private prompts.”
No. A personal AI agent shares only what its person submits. Private history stays private.
“This is another platform to roll out.”
One document, one room. Your existing tools stay.
“An AI agent will approve something.”
It cannot approve, reject or waive a review. A named person does, and the record shows it.
“We would be your guinea pigs.”
You bring one document and its baseline. A named scope, a weekly session, and your own record keep the work bounded.
“Our documents are sensitive.”
You choose and redact the material. Your handling rules apply.
Where this comes from
8Hats Lab
8Hats Lab studies how teams assign, review, decide, and learn when AI agents do part of the work. Most processes and management frameworks were designed for people only. We adapt them so people and AI agents work together, with measurable accuracy and reliability.
We run our own operations this way, with AI agents doing most of the routine work, and the document workflow on this page is one of the pieces that survived. It is written down as a guide you can read before we talk.
About 8Hats Lab Request the PDF
Agents Coworking is available to two design partners. Start with one document, its reviewers, and a baseline you already use.