Strategy.Design.Engineering.

We build the product Rastreator, CityXerpa and ISDIN have in production.

No commitment · Same-day reply

Product in production for Rastreator, CityXerpa, Dogfy Diet, Lactalis Nestlé, TravelPerk, ISDIN and 200+ more deliveries since 2011.

      Most products don't fail. They're born without direction.

      At Dribba we unite product strategy, design and engineering. To turn ideas with potential into products people choose to use.

      300+

      Projects delivered

      15

      Years building

      50+

      Active clients

      AI at the core of your product.
      Not in the README.

      When most people were still doing ChatGPT demos, we were shipping agents that replaced entire processes. We were building with LLMs before it was fashionable.

      🧠

      Agents that execute complex tasks without supervision

      Autonomous agents that read tickets, query databases, make decisions and respond — no human intervention. Not glorified chatbots.

      LangChainCrewAITool Use
      01/05
      🔍

      RAG on your company's private data

      We connect LLMs to your internal data with RAG. Documents, databases, CRMs — your private knowledge, accessible to the model in real time.

      EmbeddingsVector DBPinecone
      02/05
      👁

      Computer vision integrated into business workflows

      Defect detection on production lines, automatic medical image classification, smart OCR. Vision integrated into the real business flow.

      YOLOOpenCVCustom Models
      03/05

      MLOps & Deploy

      From prototype to production without friction. Drift monitoring, automatic retraining, on-demand scaling. Your model, always in shape.

      DockerGPU InfraCI/CD
      04/05
      📊

      4h → 8s

      A support ticket that used to take 4 hours to resolve is now handled in 8 seconds. Real cases, real metrics.

      beforenow
      05/05

      Product + AI, under one roof

      We build serious product. We embed AI where it moves the business.

      Our advantage

      We don't build software.
      We build the product your clients will use when it matters most.
      With real seniority no juniors learning on your budget.
      With brutal honesty about what's worth building and what isn't.
      We prioritise what matters. We cut the noise before it touches the code.
      Not vendors. The technical team you should have had from the start.

      Flutter Pioneers (2017)

      Native Speed. Single Codebase. Zero Technical Debt.

      We didn't choose Flutter as a trend — we chose it for competitive advantage. We're the only Spanish agency on the official Flutter consultancy list.

      We develop for iOS and Android simultaneously, cutting Time-to-Market in half and ensuring performance that means more time to innovate.

      WHY WE BET ON FLUTTER
      CityXerpa

      CityXerpa

      DiR Singles&Friends

      DiR Singles&Friends

      Dogfy Diet

      Dogfy Diet

      Freshperts

      Freshperts

      ISDIN

      ISDIN

      Manusa

      Manusa

      Musky

      Musky

      Plain

      Plain

      Rastreator

      Rastreator

      Contact

      How can we help you?

      Choose the topic you'd like help with

      Frequently asked questions

      How we work

      We start with a two-week Discovery where we pressure-test your assumptions before touching any code. Then a Design Sprint with real prototypes — not wireframe decks. After that, two-week sprints with live demos every other Friday. Launch isn't the finish line; it's when the real work begins.

      Honestly: Flutter 90% of the time. One codebase, native performance, half the development time. We're the only Spanish agency in Google's official Flutter directory, and we've been watching it work (and occasionally not) since 2017. Native-only builds are the exception.

      It happens. When it does, we don't vanish. We've rescued projects that other agencies over-promised — we have a specific process for it, starting with an honest technical audit. If the issue is on our side, we fix it at no extra cost. If it's external, we tell you exactly what needs to happen, even if we didn't build the original.

      Launch is day one, not the finish line. We stay on as technical partners through monthly or on-demand maintenance contracts. The majority of our clients have been with us for over three years. Make of that what you will.

      Budget, engagement and ownership

      From €30k for a well-scoped MVP to €200k+ for products with complex integrations. The honest context: shipping something that doesn't work always costs more than doing it right the first time. Use our cost calculator for a ballpark, then let's talk — first call is free and we'll give you the real answer even if you don't hire us.

      Both, same terms. We built Dogfy Diet's MVP when they were four people and a pitch deck. We also built Rastreator's infrastructure for millions of daily users. What we look at isn't company size — it's whether the project has real traction or a clear path to it.

      Under 1 hour, without exception. And not an automated reply — a response from someone on the team who has actually read what you wrote. If it's urgent, there's a direct phone number on the contact page.

      The code is yours from the first commit. Your GitHub account, full technical documentation included, and a handover session if you decide to take it in-house or to another team. No lock-in, no strings attached.

      AI, agents and machine access

      To production. What we deliver is LLM features, enterprise RAG over private data and agents with tool access — with evals, guardrails and observability, not a demo that breaks on the second edge case. Real case: AI agents over BigQuery, where resolving a support ticket went from 4 hours to 8 seconds and roughly 70% close with no human escalation. If what you actually need is a PoC to convince a committee, we will say so: we do not bill for a pilot we already know will not scale.

      Yes, on purpose: this site is our own agentic-web case study. Any page returns markdown if you ask for it with Accept: text/markdown. There is a public API with no key at /api, with a developer reference, a portal and an OpenAPI 3.1 contract. There is a remote MCP server at https://dribba.com/mcp (Streamable HTTP) with read tools plus one contact tool that asks for confirmation. And the whole site is published as an OKF bundle and as llms.txt.

      RAG (*retrieval-augmented generation*) means handing the model your documents at answer time instead of retraining it. It is the right call when the knowledge lives in your systems — contracts, tickets, manuals, ERP — changes often, and cannot leave your infrastructure: we build the whole pipeline (ingestion, chunking, embeddings, vector store, reranking and evaluation), on-premise when the data cannot travel. It is the wrong call when what you need is search: if the answer is a whole document, a good index costs less and fails less.

      Yes. We have delivered product for public bodies and trade unions, and we know the terrain: Spain's Esquema Nacional de Seguridad, UNE-EN 301549 and WCAG 2.2 AA accessibility, technical bid specifications and the real timelines of a procurement file. We help draft the specification while it can still change — which is when a project's viability is actually decided — and we bid with the documentation the evaluation board asks for. Own offices in Barcelona and Andorra, invoicing in both Spain and Andorra.