AI is not only changing what companies build. The companies that work this out first will operate at a scale their headcount doesn't explain. We took a point of control and turned it into the harness where an AI workforce lives and does the operational work for thousands of restaurants.

This role does the same thing inside the company: the agents that take the repetitive work off CX, Sales, Marketing, Partners, Finance and People, and the software they live in. We have the first agents in production and the first pieces of the internal platform. haddock is the AI-powered back-office for restaurants. We started as a system of record (inventories, purchases, invoices, bank reconciliation, P&L; reports) and turned it into the harness where our AI workforce lives.

That's not AI hype, it's a product running at scale for thousands of restaurants across Spain. We're now applying the same philosophy inside the company. Our vision is to lead the market through innovation, strong relationships, and by becoming the go-to tech partner for restaurant success. haddock is backed by Y Combinator, the accelerator behind Airbnb and Twitch. haddock grows fast: more restaurants, more invoices, more conversations, more deals, more partners, and with all of it, hundreds of opportunities to leverage AI every single week across CX, Sales, Marketing, Partners, Finance and People.

Rather than let every new process turn into another pair of hands, we're building an AI team focused on internal operations, designed to hyperscale the company with agents. Small enough that what you ship is visible company-wide within days. Rosetta, an internal agent that answers any question about our product or our customers' data.

It takes its context from the codebase and from production, not from the docs, finds the evidence, and explains it in plain language to non-technical people. It lives in Slack and behind our internal MCP, so anyone can reach it from wherever they already work. It never goes stale, because production is its source of truth.

Our own MCPs over third-party tools. Org-wide context,



an internal plugin carrying our curated knowledge and skills, an MCP gateway exposing scoped internal tools over OAuth, plus the rollout, the training and the adoption metrics for the entire company. Build the agents and own the software they live in.

An agent is only as useful as the harness around it: the triggers, the tools it can call, the data it can read, the service that runs it, the interface where a human reviews its work. You own all of it, from the first commit to production and to whatever breaks at 9am on a Monday. HubSpot, Zendesk, Slack, Notion, our own product databases, GCP.

Do your own project management: you find the problem, you sell the solution internally, you roll it out, you train the team that will use it, and you keep it alive. MCPs, skills and shared tooling, so every haddocker can build their own small automations on top of what you ship. 1+ years of hands‐on experience building AI systems or automations that real people use everyday. You're comfortable with product engineering.

TypeScript or Python, and you're comfortable with APIs, webhooks, jobs, queues, databases, and shipping a frontend when the workflow needs one. You're analytical and quantitative: you measure before optimizing, but you never let analysis slow down shipping. Fluent in Spanish and comfortable in English.

Most of your stakeholders work in Spanish, so this one matters. RAG systems and vector databases Deep hands‐on work with the APIs of HubSpot, Zendesk, Slack or Notion Our tech stack TypeScript, Python, langfuse, PostHog, Google Cloud Platform, mastra, Postgres, MongoDB, and several LLM providers. On top of that, the systems the company runs on (HubSpot, Zendesk, Slack and Notion) and the MCPs, skills and agents we build over them.

First and foremost, the project itself: a fast-growing startup with ambition, product-market fit, and great culture A bright modern office in Barcelona, Poblenou, probably the best office you've ever worked at At least 2 days per week at the office (in-person collaboration) + 3 days WFH for focus Technical interviews with Enric and Guillermo (Head of Tech), 2 hours.

📌 Ai Engineer - Internal Agents & Automations (Cataluña)
🏢 haddock (YC W22)
📍 Cataluña