Hosted AI agents that run without anyone prompting them.
I’m Jan Alvin Dimla. I built Autogenic — a hosted platform where a described job becomes an agent that runs on schedules, webhooks, and events, with every run logged.
01 / Platform
What Autogenic does.
One hosted runtime for agents that do recurring work: describe the job, bring your own model keys, let it run on a schedule or an event, and read the log afterwards.
- 01 Prompt to build Describe the job in plain language. A working agent comes back — steps, tools, and permissions included — ready to dry-run and deploy.
- 02 Runs on its own Cron schedules, webhooks, and platform events, with overlap protection, retries, and per-schedule run history.
- 03 Your keys, your bill OpenAI, Anthropic, OpenRouter, or any OpenAI-compatible API. Tokens bill your provider directly — no markup in between.
- 04 Every run logged Trigger, steps, tool calls, inputs, outputs, tokens, timing, status. Replay any run and see exactly what happened.
02 / How I work
Four rules the product has to keep.
Keys stay yours
BYOK isn’t a feature flag — it’s the architecture. Your keys, encrypted and scoped; your provider bill, unmarked.
Observable or it didn’t happen
If a run isn’t logged, costed, and replayable, it doesn’t ship. Agents you can’t inspect are agents you can’t trust.
Free to start
The product starts free. The run logs — not a sales deck — make the case for upgrading.
Plain language in, plain logs out
Build agents by describing them, and read exactly what they did without a PhD in observability tooling.
03 / Transparency
No investors, no markup.
Autogenic is built by one person and paid for by the people using it. There is no outside funding to answer to, no token markup in the middle, and no sales team to get past — start free and read the logs.
Keys are encrypted at rest and scoped to your workspace. Inputs, outputs, and run logs stay there, and nothing is used to train a model.
- Keys
- Encrypted at rest, scoped to your workspace, rotatable without downtime. How BYOK works
- Runs
- Isolated workspaces, permissions per agent, and every tool call logged. What a run records
- Data
- Inputs, outputs, and logs stay inside your workspace. Nothing trains a model. Privacy policy
- Billing
- Model usage bills your provider directly — no token markup, ever. Plans and limits
04 / Fit
What we are — and what we’re not.
This is
- A hosted software platform you sign up for and run yourself
- A runtime for agents that fire on cron, webhooks, and events
- Accountable to logs — every run costed, timed, and replayable
This isn’t
- An agency, a consultancy, or a scoped services engagement
- A chat product that waits for someone to type the next prompt
- A token reseller — model spend stays on your provider account
Where it runs today
Your first agent is one prompt away.
Free to start — three agents, no card, no sales call. Paid plans add agents, faster intervals, and longer run history; see plans and limits.