Most of what your organisation knows is stuck: in one person’s head, in yesterday’s chat, in a file nobody can find. Drafted is a shared surface where people and agents work the same job, so what one person works out becomes something everyone has.
A chat window is single-player: whatever your colleague worked out this morning can’t reach your session this afternoon. Drafted puts people and agents on the same surface, under a few rules neither can skip.
Before an agent may touch the work it has to search your organisation’s knowledge: your standards, your client history, the decisions you already made.
Your procedure for this kind of work: the checklist, the evidence bar, the house style. Not loaded means the write is refused.
The output lands as frames your team watches appear, compares side by side, and reviews before anyone signs anything.
What you fix goes back into the Wiki and the Skill, so it’s already fixed for every future job and everyone who runs one.
Your organisation already knows things: the standards you work to, the way this client wants their reports, the decision you made last spring and the reason for it. Today that lives in people’s heads and in old threads, and every AI session starts without it.
In Drafted that knowledge is a Wiki your agents write to and read from. Reading it isn’t a suggestion. An agent is refused its first edit until it has searched.
A written procedure changes nothing because nothing makes anyone open it. Drafted makes the machine open it. A skill attached to a project is checked before every change, and an agent that hasn’t loaded it can’t write at all.
A Skill is your method, in plain language: the steps, the checks, the standard you hold work to. Edit it once and every project, every teammate and every agent is lifted at the same moment.
Work appears on a shared surface as it’s produced, rather than in a transcript you read afterwards. Everything is a frame: documents, spreadsheets, slides, diagrams, images, live prototypes.
An agent writing from a laptop in one country lands frames on a colleague’s screen in another, instantly, with no refresh and nothing to paste.
Four options in a row take seconds to judge. The same four buried in four chat threads take an afternoon and produce a worse decision.
Every share is labelled and revocable. A whole project can never be made public, only the individual frames you choose.
There’s no Drafted chatbot to learn. It connects over MCP — the open standard assistants use to reach outside tools — to the ones your team already chose, and works the same way in all of them.
Bring your knowledge, your standards and the AI tools you already use.