Services
Web platforms, mobile apps and AI automation.
Designed and built end to end, from the first sketch to launch, with tools we’ve already tried on our own work. Tested, reviewed, and owned by you. Start with a free 15-minute call.
Software development companyBarcelona
Custom software for your business.
Web platforms, mobile apps and AI automation, built by software engineers.
We test every tool on our own projects. We only recommend what works.
Your company, connectedExample
01The filter
We test it ourselves first. We only recommend what works.
Before a tool goes into client work, we run it on our own systems: where it breaks, what it costs and what it adds. It reaches your project only if it passes, and only if it fits your case.
■ AdoptedFrom the Edge Log
Jev, from TypeSafeEach AI request now carries only the tools it needs: 8 instead of 19 for a banking question, decided in about 249 ms. In a client’s system 2 days after we measured it, under a week from its release.Site down. Fixed before we knew.Monitoring caught a client site going down and handed it to an AI agent. It was back up within 9 minutes, before anyone on the team had looked at it.All business knowledge in one place, ready for AIA company’s documentation, data and processes, structured in one place so AI can use them for personalised outputs: answers, quotes, follow-ups. Running on our platform; live for a client in November.A legal assistant that cites Spanish lawAnswers Spanish legal and tax questions from official sources only: BOE law, all 17 autonomous communities and 41,237 binding tax rulings. Every claim carries its citation. Client access is next.
× DroppedAnd why
Adversarial multi-agent debateWe set agents to critique each other’s plans and code in loops. They converge on agreement too fast, and a strong model debating a weaker one ends worse than the strong model’s first answer. Now each decision goes to the most capable model.
A general AI model for every small decisionMatching bank movements and suggesting accounts went to a general model that answered in free text, for decisions that only needed a choice from a list. We moved them to Jev, which scores the options and always answers with one of them.
Claude Managed Agents for our fixing agentsWe set it up to compare with the agents that fix our production errors. We removed it the same day: it bills per API token, and our agents run on fixed subscriptions so their cost stays predictable.
A queue-based ticketing systemRequests waited in a queue until someone picked them up. We replaced it with a support layer connected to every channel, where agents investigate first and an engineer joins the same thread when needed.
02What we build
Systems that help businesses grow.
On our bench
AI-run media and marketing.
We’re testing it on SEIRO first. It becomes a service when it brings leads that turn into clients, and pays for itself.
TestingNot for sale yet. Follow the test
03How we build
Generating code is easy. Making it hold up in production is engineering.
We design the architecture, the data and the integrations, and decide what’s worth building. AI makes us faster; tests and code review on every change make sure nothing slips through.
How a build runs · illustrationTests and review from the start
Designed by engineers
Senior engineers decide the architecture, the data and the security. AI works from that plan.
Checked on every change
Automated tests, AI review and a person on every change, from day one.
Faster
A logistics platform rebuilt in six months by three people. Without AI it would have taken years.
Yours
You own the code. Nothing depends on a tool we can’t replace.
04Work
Recent work, running and launching.
The platform we run SEIRO on
It started as an invoice template. Now it’s where every tool gets tested first.
Delivery Mates’ platform, rebuilt in six months
From extra developer to leading the rebuild of Delivery Mates’ whole platform.
A sales system that follows up on its own
For a European manufacturer that ran sales on spreadsheets.
A private members’ app in four weeks
An invitation-only platform, already live: redesigned, its security reinforced, and built as a cross-platform iOS and Android app.
05Edge Log
What we’re trying, keeping and dropping.
Each entry shows how far a tool got on our own work. It reaches clients only when it’s ready for them.
Sept 2026AdoptedJev, from TypeSafe Each AI request now carries only the tools it needs: 8 instead of 19 for a banking question, decided in about 249 ms. In a client’s system 2 days after we measured it, under a week from its release.Ready for clientsAug 2026AdoptedSite down. Fixed before we knew. Monitoring caught a client site going down and handed it to an AI agent. It was back up within 9 minutes, before anyone on the team had looked at it.Ready for clientsMay 2026AdoptedAll business knowledge in one place, ready for AI A company’s documentation, data and processes, structured in one place so AI can use them for personalised outputs: answers, quotes, follow-ups. Running on our platform; live for a client in November.Ready for clientsAug 2026AdoptedA legal assistant that cites Spanish law Answers Spanish legal and tax questions from official sources only: BOE law, all 17 autonomous communities and 41,237 binding tax rulings. Every claim carries its citation. Client access is next.Works for usDec 2025AdoptedAutomating our financial systems Agents read receipts in any format and from any country, match them to bank movements and book them only when every check passes. What needs a person goes to one review queue.Works for usSept 2026AdoptedA faster way for agents to test our apps Our internal testing tool: it reads each screen as text instead of sending screenshots to a large model, so an agent walks a whole app in minutes. 56 screens in 25 minutes on a real app.Works for us
Real entries from our own work. Read the full log →
06Contact
Tell us what you want to build.
Three lines is enough. We reply with questions or a first idea of how we’d build it. If we’re not the right fit, we’ll say so.