Every service that talks to an AI needs the same things: a queue, retries, rate limiting, history.
oqueue takes care of all of it so you can focus on what your service actually does.
First in, first out
Jobs are processed in the order they arrive. No request jumps the queue,
and a rate limit keeps your AI server from getting swamped.
Retries on failure
If the AI server hiccups, oqueue tries again automatically up to 3 times.
Only after all attempts fail does the job get marked as failed.
Conversations that remember
Group jobs into a conversation thread and oqueue stitches together
the full message history on each turn, so your AI always has the context.
One key per service
Each service authenticates with its own API key and gets a tailored system prompt
and knowledge base, injected into every request without any extra work.
See what is going on
A clean admin panel shows live job counts, lets you browse and filter the queue,
read conversation threads, and manage API keys.
Works with your AI server
Point it at any Ollama-compatible endpoint and you are done.
No platform lock-in, no surprise bills.