AI, apps and algorithms
made to measure.
We understand the problem. We build it. We put it in production.
A small, senior team trained in Mathematics and Computer Science. AI systems on your data and documents, apps and MVPs, and custom algorithms. Per project or embedded in your team: you talk to the people writing the code.
WHAT ARE YOU WORKING ON?
WHERE WE'D START
We'd start with your data and documents: which model and architecture fit (RAG, agents or classifiers) and how we evaluate it on real cases before putting it in production.
We'd start by scoping the first version worth building. We deliver it in milestones you test yourself, wherever your users are: web, iOS or Android.
We'd start by mapping the process exactly as it runs today, digitizing the documents and extracting the data. You get a dashboard to watch it run, hands-free.
We'd start by formalizing the problem with mathematical rigor and a Python prototype on your data that validates the approach. Then we integrate it into your system, measured and documented.
We reply within 48 business hours. No strings.
01 — WHO WE ARE
AI and software
built on maths.
We're a small, senior team. We studied Mathematics and Computer Science and have spent years shipping product to production, with real users: management platforms, consumer apps, RAG systems to chat with documents, AI-generated-text detectors and automatic document digitization. No unnecessary layers: you talk to the people writing the code, we deliver by milestone and we plug into your team when needed.
- 01 Fixed price before the first line of code.
- 02 A release every week, from week one.
- 03 Code in your name from day one. Any team can inherit it without fear.
02 — PROCESS
From problem to production.
From the first call to a system in production. Every phase ends with something you keep.
The shortest path
-
DISCOVERY
We listen to the problem and look at your data and context. Together we decide what's worth building and what isn't. You keep: written scope, dated milestones and the fixed price. If you don't want to continue, the plan is yours.
-
DEFINITION
Architecture chosen (and the model, if AI is involved), key screens if there's an interface, and a dated milestone calendar. This is where we decide the few things the first version must do and which ones can wait. You keep: the milestone-by-milestone plan and, where it applies, a clickable prototype.
-
BUILD
Every week you get a new version in an environment where you test it, and a 30-minute call to walk through it. If there's AI, we evaluate it on your real cases, not on demos. You see progress, you decide, we correct course before it gets expensive.
-
DELIVERY
System in production, repository in your name and documentation so another team can carry on without calling us. You get the keys, not a demo.
If you want to keep going, we continue with the same team that built it: by milestone or embedded in yours. If not, you keep the code and what you learned, and pay only for the milestones closed.
03 — PRICING
Fixed price by milestone, before we start.
Closed projects by milestone; team reinforcements by sprint or hour bank. Either way, in writing before we start. If the scope changes, we change it in writing, not on the invoice.
-
WEEK 0
Discovery and scope
Fixed price in the proposal
Five days with you. You keep the scope document, the milestones and the fixed price. Deducted if you continue.
-
PROJECT BY MILESTONE
Week 0 included
Fixed price in the proposal
An AI system, an app or a custom algorithm, in production on your domain. Dated milestones at a fixed price: you validate each delivery before paying for it.
-
EXTENDED TEAM
Reinforcement or continuity
Sprint price in the proposal
The same team, embedded in yours: two-week sprints at a fixed price or an hour bank. No lock-in.
Always included
- Fixed price in writing before the first line of code.
- A release every week, from week one.
- Repository in your name from day one. Test and production environments in your accounts.
- Documentation so another team can carry on without calling us.
- On closed projects, no hourly billing: price by milestone. No account manager.
Rules of the road
- 01The scope document lists features, screens or target metrics one by one, and also what stays out. Anything that appears in neither list is resolved inside the milestone, not on the invoice.
- 02If the scope changes, we change it in writing with its impact on date and price, before we continue. Never on the invoice.
- 03A delay is ours whenever it doesn't come from a scope change you signed or from something you owe us for more than 5 business days. If the delay is ours, it costs nothing more.
- 04You can stop at the close of any milestone. You pay the milestones closed so far and keep everything built, with the repository and accounts already in your name.
- 05You bring one decision-maker and one hour a week to test the release. We bring a single point of contact from Week 0 to handover, replying the same business day.
The price is fixed at the end of Week 0 and doesn't move after that, unless you change the scope.
Request a proposal04 — PORTFOLIO
What we've built and is in production.
Platforms, apps and AI systems we've taken to production, plus an optimization side project that shows how we think.
Optimal routes
SIDE PROJECT · PYTHON05 — WHO IT'S FOR
Who it's for. And who it isn't.
FOR YOU IF
- You have a concrete problem, with data or documents to work on, and you want a system in production, not a demo.
- You have a product idea and want to show something real to users and investors, not a slide deck.
- Your tech team is missing one piece, AI, backend, an app or an algorithm, and you want it built by people who have already put it in production.
- You can spend one hour a week testing what we ship and deciding.
NOT FOR YOU IF
- You want the cheapest app. Others build it for less; we build the one that holds up when users arrive.
- You want thirty features in the first version. A tight first version is the shortest path; thirty features is the longest.
- You want a report or a talk about AI. We build.
06 — STACK
The problem picks the stack.
Flutter and Firebase to launch fast on iOS, Android and web from one codebase. Python for AI, data and optimization. Node and Google Cloud behind it. Whatever it takes in between. We pick from this depending on what your problem needs.
We pick tools for what your product needs, not for fashion. And so another team can carry on without calling us.
07 — QUESTIONS
Frequent questions before you write to us.
01 How much does a project cost?
Fixed price at the end of Week 0, in writing. What moves the price: scope, integrations, data volume and whether it goes to the app stores. What doesn't: changing your mind in Week 0, that's what it's for.
02 How long does it take?
It depends on the scope, and that's what Week 0 is for: when it ends you have a date and a price in writing. A tight first version is measured in weeks, not months, with a release every week from week one.
03 What kind of AI projects do you do?
Systems that work on your data and documents: RAG to query your documentation, agents that execute tasks, classifiers and detectors (for example, of AI-generated text) and automatic document digitization. Always evaluated on real cases before shipping.
04 I know nothing about tech. Is that a problem?
No. You bring the problem and your knowledge of the business. We choose the technology and explain every decision in plain language.
05 What if the AI isn't good enough?
Nobody can guarantee that before testing it on your data, and anyone who does is lying. What we can: we measure it early, on your real cases and agreed metrics, and you know in weeks, not in a year. You keep the code, the evaluations and the data to decide with numbers. You can stop at the close of any milestone and pay only for the milestones closed.
06 And if it works?
We continue with the same team that built it, by milestone or embedded in yours. Or you keep it and hire whoever you want: the code doesn't know who wrote it.
07 Can you join my team instead of running the whole project?
Yes. We work on closed projects or as a reinforcement for your team: two-week sprints or an hour bank, with your tools and your repository. We decide on the first call.
Software and AI studio in Spain, working remotely
We're a studio, not an agency with account managers nor a freelancer working alone. All of Spain, remotely.