AI Chatbot

ai-chatbot

Vercel's Next.js AI Chatbot — AI SDK 7 streaming chat with tools and artifacts, Auth.js with a guest flow, Drizzle over Postgres, image attachments — running on Cloudflare Workers + Static Assets through @opennextjs/cloudflare, with its models on Workers AI through the Worker's own binding, its Postgres a Supabase project reached through Hyperdrive, its attachments in an R2 bucket, and the project, the Hyperdrive config, the bucket and the schema all provisioned by Terraform and Orun's db-migrate through Orun, the way the Lumen baseline does it — and packaged as an Orun baseline: one command takes a new chatbot from nothing to live on stage and prod, in its own repository, under its own name, in ONE landing.

The template's code is the code. What the port changes is what the Workers runtime and a fresh, unattended deployment force: the Postgres client comes from the HYPERDRIVE binding, one per call; attachments go to the BLOB R2 binding and are served back at /api/files/<key>; the Vercel-only modules — Blob, BotID, @vercel/functions, @vercel/otel, @vercel/analytics — are each replaced by their Cloudflare counterpart or a no-op; the models default to Workers AI (so a fresh product chats with no model key, and the AI Gateway's models appear only once its key is set); and the session key defaults to one the database generates for itself (so no secret has to be set by hand). What the baseline adds is everything needed to make a copy of it somebody else's — built the way cirrus-baseline builds a product, in one landing. See specs/epics/ai-chatbot for the design, the milestones and the as-built status, including every departure.

Instantiating a product

git clone --depth 1 --branch <baseline-tag> https://github.com/sourceplane/ai-chatbot
cd ai-chatbot
orun new --blueprint repo-blueprint.yaml --out ~/products/acme-chat --run-hooks --resume \
  --set reponame=acme-chat --set productname="Acme Chat" \
  --set githuborg=acme-inc --set subdomain=<workers-dev-subdomain>

Or, once the platform's registry lists it, from the Orun console (Baselines → ai-chatbot → Build). Two phases: 01-product creates the repository, mints the six brokered credentials (Cloudflare and Supabase) and lands the whole product as one pull request, whose convergence on main provisions and deploys it per environment in a fixed order —

supabase ──▶ cloudflare-hyperdrive ──▶ db-migrate ──▶ web
                                  cloudflare-r2 ──▶ web

— each edge mode: enforced on a push to main and advisory on a pull request; then 02-docs records the live deployment from probed facts. BOOTSTRAP.md is the operator guide; docs/phases/ has a page per phase.

The new repository receives the product only — the app, the Terraform roots, the migrations, the tests, its CI, its tooling, a README of its own and its operating contract — renamed throughout to the product's name, repository, Worker, Supabase project, Hyperdrive config and bucket. None of this repository's machinery ships. No AI vendor key is asked for: the platform brokers none, and the chat runs on Workers AI; the AI Gateway's key is the owner's to add afterwards (the product's README, "Models"). The Supabase organisation the build uses needs room for two projects, stage and prod — the free plan's limit.

What is here (the product)

Getting started

pnpm install
pnpm run wire:fixture   # re-render apps/web/wrangler.jsonc from its template (committed; this changes nothing)
pnpm typecheck
pnpm lint               # the template's own biome/ultracite in apps/web; eslint in the packages
pnpm test               # tests/web, tests/db
pnpm build              # next build + opennextjs-cloudflare build → apps/web/.open-next

The build needs no database, no Cloudflare account and no model key. Running it locally — a Postgres, pnpm --filter @app/web db:migrate, next dev or wrangler dev (Workers AI is reached remotely, so wrangler login first) — is in the product's README (.baseline/product/README.md), which is what a product built from here carries.

How it maps to Vercel

CI