PromptQL reframes AI assistance as a team sport. Rather than a single person chatting with an assistant, you spin up a bot for a piece of work — a reliability dashboard, a data quality sheet, an API spec, a monthly metrics deck, a delivery board — and invite the colleagues who actually hold the context. When the bot hits a gap, it asks the right person; when it finds an answer, it writes it back into the artifact next to the discussion.
The product's framing is that three people usually know where to look, and the bot checks all three. In its own examples, three people know what the data does not say, what the code does not say, and what the charts do not say. The bot's job is to reach those people, test their theories against the source systems, and keep the shared artifact correct rather than duplicating knowledge into yet another silo.
Technically, PromptQL bots can run continuously, work on an isolated computer with a browser, build their own tools for arbitrary APIs or databases, and create artifacts and apps. The multiplayer layer includes per-user data isolation — a user cannot ask the bot to access another user's data — editable and deletable messages in shared chat, and private bots users can spawn for sensitive work. An automatically maintained knowledge base captures what the team learns so context compounds rather than resets.
PromptQL also publishes detailed positions on shared context architecture, security, enterprise deployment and pricing, and ships an SDK for teams that want to build multiplayer AI agents themselves.
