We are building the runtime that AI agents run on in production.

What we are doing

Everyone is building AI agents. Almost nobody has worked out how to run them properly: versioned, observable, governed, and portable across models and clouds. That gap is the company.

Stackbone is the control plane for agents in production. The agents run in the customer’s own infrastructure, with the customer’s own model keys. We are model-agnostic and cloud-agnostic on purpose, because the alternative is asking a bank to hand its data to a vendor, and banks say no.

How we work

We are a small distributed team between Madrid and Montreal. Small enough that what you build ships the same week and you see who uses it.

A few things that are true about working here:

• You will talk to customers. Engineers included. We do not have a layer of people between the team and the people using the product.
• Decisions get written down. Distributed across time zones means asynchronous by default, which means writing things down properly.
• Scope is wide. At this size nobody has a narrow job description. If you want clear boundaries and a defined lane, this is the wrong stage.
• Two languages. The team works in Spanish and English. You need to be comfortable in English; Spanish is useful, not required.

Our stack

TypeScript across the board, Node, Docker for everything deployable, Postgres and Redis, S3-compatible storage. Monorepo on GitHub with Actions for continuous integration. Linear and Slack.

Open roles

We do not have a formal opening right now.

That does not mean we are not hiring. At this stage the right person usually arrives before the job description does. If you read the above and thought you would be good at it, write to us.

Send a note to [email protected]. Tell us what you would want to work on and point us at something you have built. Skip the cover letter.

Working with us without joining us

If you run an agency or a consultancy and want to build on Stackbone for your clients rather than work for us, that is a different conversation and we would like to have it. Same address: [email protected].