Back to home
|

Custom software with AI

Agents, automations and apps that reach production and get used. End to end, with AI only where it helps.

Talk about the project
Auri, Tesora's AI assistant, working out whether paying a loan off early is worth itMimify live: 40 agents testing a website

CTO of Tesora: AI finance on web, iOS and Android.

From idea to production, no demos left in a drawer.

Many AI projects end as an impressive demo that nobody uses. The challenge is making it work with your data, your tools and real users, at a reasonable cost. That's where I work.

Who it's for. Companies that want to automate a process, put an assistant on top of their data or launch an AI product. And engineering teams that need help taking it to production.

What I build. Agents connected to your tools, automations, search and assistants over your documentation, and complete web and mobile applications.

How I work. First I check whether AI helps in your case. Then I prototype with your real data and build with frequent deliveries you can try.

How to start. Tell me the problem you want to solve. After a call I'll give you a plan with phases, timelines and budget.

What I can build

AI agents

Assistants that query your systems, run calculations and act, connected through MCP and tools, with evaluation to know they get it right.

Auri, Tesora's agent, answering with real banking data
Auri, at Tesora.

Process automation

Repetitive tasks turned into automatic flows: documents, email, data and operations, with AI where it helps.

Assistants over your data

Search and assistants that answer from your documentation and data (RAG), citing sources and with quality control.

Web and mobile apps

Complete products with React, Next.js, React Native, Node.js and PostgreSQL, ready for real users.

Tesora, on web, iOS and Android.

Prototypes and MVPs

A first working version to validate the idea fast before investing big.

X-Ray, a financial health engine built at HackSpain 2026
X-Ray, at a hackathon.

How we work

  1. Discovery

    The problem, the data and whether AI is the right tool.

  2. Prototype

    A minimal version that proves it works with your data.

  3. Iterative development

    Frequent deliveries you can try, not a black box for months.

  4. Production

    Deployment, metrics, evals and model cost control.

Already in production

  • Stack
  • React and Next.js
  • React Native
  • Node.js and NestJS
  • PostgreSQL
  • OpenAI, Anthropic and Gemini
  • MCP and RAG

FAQ

How much does it cost and how long does it take?

It depends on scope. After a first call I give you a plan with phases, timelines and budget. Starting with a focused prototype lowers the risk.

Which AI models do you work with?

OpenAI, Anthropic and Google models, among others. They're chosen for quality, cost and privacy in each case, with no vendor lock-in.

What happens to our data?

It's decided upfront: which provider is used, where data is processed and what gets stored. Privacy is a design requirement, not an add-on.

Can you work with our engineering team?

Yes. I can build the whole product or work alongside your team, and combine it with training so they can maintain it afterwards.

Your idea, in production

Tell me the problem you want to solve. I'll tell you whether AI makes sense and how I'd build it. Or email me at hugoliva158@gmail.com

AI training for companies Talks, workshops and in-company programs so your team uses and builds with AI: agents, Claude Code, Codex, MCP and n8n.