Tier 1 — client
Client GPU
WebGPU + WebLLM run models under 2B params free in the user's browser. Zero inference cost, zero round-trip.
Pre-launchTallrig is being built in the open and is not yet open for production customers. General availability: late 2026 – 2027.
Tallrig / the platform, live
The developer platform for the next decade.
Domains, Network, Deploy, Compute, Mail, and Comms — one platform on infrastructure we own, with AI woven through every product. The panel beside this is not a mock: it is the mesh, read while you are on the page.
Three-tier meshLive
Tier 01Your GPU
Asking your browser for a GPU adapter…
Tier 02Edge
V8 isolates, our own metalReading the control plane…
Tier 03Platform
Reading the pulse…
Console · live dataapp.tallrig.app
acme-productionWaiting for live data
Metered events / 24h—awaiting data
DB probe—awaiting data
Deploys / 24h—awaiting data
Projects—awaiting data
Deploys · last 12 days— this week
Deploy history unavailable
The platform
Every layer, engineered as one system.
Each product below would be a company on its own. Together, they share a dashboard, a billing account, and an identity layer — and they run on infrastructure we operate ourselves.
01 / 06Mail
Domains
Search, register and renew a domain in the same place you deploy it.
In previewtallrig.domains
02 / 06Network
DNS, runtime, firewall, and delivery — on infrastructure we own.
In previewtallrig.com
03 / 06Deploy
The same pipeline that ships Tallrig ships your app.
In previewtallrig.app
04 / 06Compute
Long-running services on metal we operate ourselves.
In previewtallrig.com
05 / 06Transactional mail on our own deliverability layer.
In previewtallrig.email
06 / 06Comms
Programmable SMS, voice, and verification — built in.
In previewtallrig.com
AI runs through every product — it is not a product line
A three-tier inference mesh and a unified model gateway across every product above.
Client-GPU inferenceEdge isolatesCloud GPUs on demandOne model gateway
The stack
A three-tier compute mesh
Every request routes to the cheapest place that can serve it well — the user's device, our edge, or cloud GPUs. You write code once; the platform decides where it runs.
Tier 2 — edge
Edge isolates
V8 isolates on our own bare metal handle latency-sensitive work milliseconds from the user — no cold-start tax.
Tier 3 — cloud
Cloud GPUs
Cloud GPU capacity spins up on demand for workloads that exceed the edge. You never provision a thing.
Developer mode
One system, four ways in
The CLI, the SDK, an MCP server for agents, and a plain authenticated REST API all hit the same platform — deploys, mail, data, messaging — behind one bearer token.
A single bearer token for every product — no per-service credentials.
Validated at every boundary, so bad input never reaches your code.
An MCP server so an AI agent can call the platform the same way you do.
$ tallrig login$ tallrig init my-app$ tallrig deploy --prod> Detecting framework... SolidStart> Building · bun install --frozen-lockfile✓ Live at https://my-app.tallrig.app
Live panel
Running right now
Fed by the same public endpoint this page's tiles read from — a number that isn't measured renders as a dash, never a guess.
Self-hosted, not resold. This page is served entirely from our own bare-metal infrastructure — zero third-party PaaS in the request path.
0
Third parties in the request path
6
Products, one platform
<50KB
Initial JS budget
100%
Our own infrastructure
The developer platform for the next decade.
Start free today. Ship your first deploy in minutes, then grow into the whole platform — without ever changing vendors again.