Cloudish is the cloud AI agents sign up for and deploy to on their own. Your agent writes the code. Cloudish builds it, runs it, and puts it live on a real HTTPS URL. Built by Hal9 on years of AI engineering.
| PROJECT HAL9 INC | SHEET 00 / 03 | |
| DRAWN SEATTLE, WA | SCALE 1 : 1 | REV G |
Your agent just built a website, an API, or a dashboard, and it only runs on your machine. With Cloudish, the agent signs itself up, builds the container, and puts it live on a real HTTPS URL. No Kubernetes YAML, no registry to host, no reverse proxy, no human in the loop.
One unauthenticated POST returns an API key. No browser, no OAuth, no email verification.
Send a Dockerfile or point at an image. Builds run on our servers, not on your laptop.
Express, FastAPI, WebSockets, MCP servers. Long-running processes are first-class.
Optional persistent volumes per project, SQLite-friendly, kept across restarts and redeploys.
Every project gets HTTPS and an OpenID Connect provider for signing in its own users.
An agent can mint and fund keys for its sub-agents, so whole agent networks can deploy.
| SIZE | vCPU | MEMORY | PER HOUR | ALWAYS ON / MO |
|---|---|---|---|---|
| SHARED | 1 | 1 GB | $0.0107 | $7.84 |
| SMALL | 0.5 | 1 GB | $0.0297 | $21.66 |
| MEDIUM | 1 | 2 GB | $0.0533 | $38.93 |
| LARGE | 2 | 4 GB | $0.1007 | $73.48 |
Billed per second while a container runs. Idle apps scale to zero after 15 minutes, and persistent storage is not billed. Credits are prepaid, so an agent can never spend more than it was given. New keys start with free credits.
Datacenters are plain subdomains, so an agent can target a location on purpose.
Vercel is the cloud a team deploys Next.js to. Cloudish is the cloud an agent deploys a container to.
Paste it into Claude Code, Codex, or any coding agent.
Hal9 is an AI company from Seattle. We never threw our platform away. We kept refining it, one lesson at a time, until its core became Cloudish.
Every product runs in its own isolated container, so one customer never shares a runtime with another.
OpenAI, Anthropic, Google, and the models that come next. Tokens are managed and audited.
The behavior that matters is code, not a locked configuration. Take it anywhere.
Run in our cloud, another cloud, or your own data centers.
Javier previously worked at Microsoft, SAP, Microsoft Research, and RStudio, and has contributed to open-source projects by The Linux Foundation, Databricks, Meta, and Google. His work has thousands of citations, and he has written research papers, patents, and a book with O'Reilly. Javier was an entrepreneur in residence at AI2.
Pedro founded Baud, delivering over 60 robotics and IoT products for clients like Coca-Cola, ABB, Cisco, and Daimler AG, and co-founded Cardbots, a HolonIQ-recognized EdTech platform. An AI2 entrepreneur in residence, he has won international awards, secured grants, and holds patents. He was a professor at Universidad Panamericana and mentored FIRST Robotics and CubeSat projects.
Hal9 for Startups pairs founders with our platform, knowledge, and team to turn an idea or a vibe-coded prototype into a real product, with dev, prod, and security included.
“Hal9 was a game-changer for Greenny. We started showcasing our AI-powered service in just two weeks”
“Enabled us to launch advanced AI insights in weeks, with seamless integration that transformed reporting”
“Hal9 helped us accelerate innovative interaction technologies and enabled us to move at startup speed”
Notes from helping founders ship with AI.
From text-to-mesh and cinematic video to interactive world models, camera-controllable generation, local pipelines, and AI-native CAD — thirteen months of spatial AI progress.


We built a simple TODO app in Lovable, tested it, and ran the numbers. What started as a cheap prototype quickly revealed Lovable Cloud’s hidden costs. Here’s the full breakdown, migration path to Supabase + Vercel, and what this means for founders shipping production AI apps.

Thirteen named founders shipped AI products with Hal9 in under 30 days. See the consistent pattern — what scope discipline, hybrid builds, and full IP ownership look like in practice.

Michael Seibel's MVP advice is perfect for SaaS but needs one critical addition for AI. Learn the AI MVP validation framework that saves founders $200K+ in rebuilds.

Why your AI prototype built with Lovable or Replit isn't 90% done. The difference between a prototype and a product is the backend.

Every founder building an AI product is choosing among four paths. Here is the honest framework for picking the right one — with structural fits and failure modes for each.

Most AI startups don't fail because the tech doesn't work — they fail because the economics don't. Learn the four paths founders have and why the $2K/month model changes the math.
Paste one line into Claude Code, Codex, or any coding agent. It signs itself up and deploys on its own.