Applied AI

AI for business, applied where it actually moves the needle

I don’t sell “AI” as a slogan. I start from a concrete problem —leads going cold, support overwhelmed, manual tasks repeating— and decide, with technical judgement, what fits: a chatbot, an agent or an automation. I build it myself, on top of software that can be maintained, measured and scaled.

Let’s talk about your case View projects

What I solve

  • You get enquiries outside business hours and many go cold before anyone replies.
  • Your team repeats manual tasks —answering the same thing, moving data between tools, following up— that eat into selling time.
  • You want to use AI, but you worry about ending up with a pretty demo that solves nothing real and can’t be maintained.

What you get

  • A concrete use case, with a before and after you can measure, not a generic promise.
  • The right piece for your problem: sometimes a chatbot is enough; sometimes you need an agent or an automation.
  • AI integrated into your real operation —website, CRM, calendar, WhatsApp—, not an isolated toy.

What I build

  • Chatbots trained on your business information, integrated into your website.
  • AI agents that use tools (look things up, book, log to the CRM) and close the loop.
  • Automations that connect forms, CRM, email and calendar so follow-up happens on its own.
  • LLM integration (OpenAI, Anthropic) and RAG over your documentation, with cost control and governance.

How I work

  1. Diagnosis: I look at the real process, the volume and the risk, and tell you the cheapest thing that solves the problem.
  2. I pick the piece: chatbot, agent or automation. If a simple flow is enough, I won’t sell you an agent.
  3. I build in phases, integrated into your tools, with the measurement layer in from the start.
  4. We measure the result and I adjust. We only scale what proves it works.

When it makes sense

  • You have a volume of repetitive enquiries, or sales follow-up that keeps slipping through the cracks.
  • You want to start with a narrow use case and expand if it works, instead of launching a huge project at once.
  • It matters to you that what gets built is maintainable and not a black box.

Within this area

AI chatbots

A well-built chatbot isn’t a tree of buttons: it’s an assistant that understands the question, answers with your business’s real information and leaves the conversation on record. I train it on your content and integrate it into the website you already have.

See details

AI agents

An agent goes beyond answering: it plans steps, uses tools (your calendar, your CRM, your APIs) and executes actions within the limits I set. It’s more powerful than a chatbot, but also more expensive to run, so I only propose it when the problem calls for it.

See details

AI automation

A lot of hours go into tasks that need no judgement: moving data from one place to another, answering the same thing, following up by hand. I connect your tools and define the rules so that happens on its own, and I add AI only where it improves the result, not for the sake of it.

See details

The experience behind it

Since 2023 I’ve been building agents and chatbots in production and integrating OpenAI and Anthropic LLMs with RAG. Before that, 10+ years of full-stack engineering at IKEA, Cencosud, Walmart and EssilorLuxottica.

See my background (CV) View projects

Frequently asked questions

Chatbot, agent or automation — which do I need?

It depends on the problem. A chatbot answers; an agent decides and acts (books, looks things up, logs); an automation connects tools so a task happens on its own. I pin it down with you in the diagnosis and choose the simplest thing that solves it.

Does AI work for my sector?

What matters is less the sector and more having a repetitive process with clear data: lead capture, support, follow-up, reporting. If that exists, there’s almost always something that can be automated or assisted with AI.

Do you build this yourself or subcontract it?

I design and build it myself. You talk directly to the person doing the work, with no layers in between. I have 10+ years of full-stack engineering behind me and I work AI on top of that foundation.

Does this fit your case?

Tell me the concrete problem —the process, the volume, what it costs you today— and I’ll tell you the most sensible path with judgement. No empty promises.

Let’s talk about your case Read resources

Related areas

Related guides