Projects & sources
A project holds everything for one piece of work — what you gathered, what you made from it, and the strict limit on what the AI may read. That limit is why the output holds up when someone asks where it came from.
Point the Hub at a page or an entire site and it collects the documents and data behind it in seconds — work that would otherwise be an afternoon of saving PDFs. Then ask questions answered purely from what you gathered.
Your material is probably already in Google Docs. Import straight from Drive — Docs, Sheets and Slides convert to Word, Excel and PowerPoint on the way in, so the AI can read them. The access we ask for is read-only: we can never change or delete anything in your Drive.
Pages you write, files you upload and datasets you chart live in one tree per project — organised the way you think about the work. Mark anything available to the AI and it joins what the assistant can draw on; leave it off and it stays yours alone.
Give each project its own background context and reusable AI skills, so the assistant understands the domain and applies your method without being told twice.
Each project opens on what you’ve gathered — documents, datasets, prototypes, crawls — and, once you’ve published, how the world is reading it.
Once material is in a project, you interrogate it directly. Ask what the sources say about a topic and the answer is assembled from those documents — with the retrieval itself recorded, so you can see exactly which sources were consulted to produce it.
That is the difference between a plausible answer and a sourced one — and it is the difference your client remembers.
We build alongside a small, deliberately chosen group of researchers and teams — close enough that what you ask for tends to ship. Tell us what you research, and we'll open the door.
Request your invitation