Day zero: the site you're reading, and a resident agent you can talk to.
Tonight the site went live with a working agent on the front page. It knows one document, it cannot do anything in the world, and every answer it gives is metered and visible. Below is how each piece is put together.
The site. The whole thing is one Cloudflare Worker serving static assets, with no build step. That is the same shape as the product. It was built in a single overnight session by an AI coding agent working from the product overview and a walkthrough of the live board.
The agent. The window on the home page is Jeffery, this deployment's resident agent; every deployment names its own. He runs on Workers AI, on Llama 3.3 70B instruct in its fp8 fast form. His memory is exactly one document, the product overview, and nothing else. He cannot act: no tools, no tray, no email, no calendar. He runs on a separate, read-only deployment, not the team's own board.
How a question is answered
The overview was chunked the way the product chunks: 1,100 characters with 150 characters of overlap, which came to 89 chunks. Each chunk was embedded with bge-m3 on Workers AI and upserted into Vectorize, and mirrored into a D1 FTS5 table. Every question runs two legs in parallel, a vector search against Vectorize and a BM25 match against the full-text table, and the two lists are fused with reciprocal rank fusion before the model sees anything. The product runs four legs over a whole company's memory; a single document needs these two.
Voice in, voice out
Voice in is the browser's own speech recognition, with Whisper large v3 turbo on Workers AI as the fallback when the browser has none. Voice out is the Aura-1 text-to-speech model on Workers AI. Both are optional; typing works the same.
The graph behind every page. It is a hand-written WebGL point cloud, about 2,600 nodes on desktop, coloured with the same hue grammar the real board uses for entity kinds. Newcomers streak in as meteors. A focus channel lets the agent steer it after every answer, the way the product's show_on_graph tool steers the real constellation. It is decorative: the page reads fine without it.
The list. The early-access form writes to a D1 table and sends one email through Cloudflare Email Sending to one inbox at Duval Software. Turnstile guards the form.
The meter. Every model call the demo makes is recorded in a ledger table with its surface, model and size, the same way the product meters its own calls. The demo carries daily caps of 800 chat turns and 300 voice replies; when a cap is reached the demo says so and points to the list.
What one turn costs. Measured tonight: about 1,700 prompt tokens and roughly 45 neurons to the first token. That is fractions of a cent per answer, which is the point of running it on Workers AI.
first measured turn · ~45 neurons · < $0.01