Hodman AI
How Hodman works
From runtime choice to a live app.
Hodman is a product loop: choose where the agent runs, define the goal, inspect preview, publish when ready and keep improving with context.
Desktop local or Hodman cloud
Your ChatGPT subscription
Managed model usage
Preview, publish, domain and iteration stay in one workflow.
01
Choose where Hodman works
Run locally in Hodman Desktop or use Hodman servers. Cloud runtime is easier for hosted work and teams, but servers cost money, so we ask for payment up front.
Then choose how AI is paid for: connect your own ChatGPT subscription, or use managed model access through OpenAI API where usage is billed through Hodman.
02
Create a project and set the goal
Pick the project type that fits best: landing page, internal tool, assistant, dashboard, automation or API-first app.
You describe the outcome, audience, constraints and what “done” should look like. Hodman turns that into a working product direction.
03
Hodman asks, builds and opens preview
The agent clarifies missing details, writes code, runs the app and shows the result in preview so you can judge behavior, not just files.
Preview, logs, database state and project files become the feedback loop for the next fix or improvement.
04
Use preview and published versions separately
Keep a preview version for changes and a published version for users, or skip publishing and share preview directly while you iterate.
This lets you test before release without forcing every experiment into production.
05
Assign a domain or subdomain
When the app is ready, connect your own domain or subdomain and make the project reachable as a normal web product.
Hodman keeps the product loop open after launch, so you can keep fixing, measuring and improving.
Start with the runtime that matches your work.
Use Desktop for local experiments or create a cloud account when you need hosted runtime and team access.