Services

AI built into the product, not bolted onto the corner of the screen

Search that understands what someone meant, forms that fill in what can be derived, assistance that walks a user through the work — rather than a chat window nobody opens twice.

What it means

The most common shape of "AI in a company" today is a chatbot glued into the corner of an existing site. A user tries it once, gets a generic answer, and never comes back.

It gets interesting when the AI disappears into the product itself: search finds what a person meant even when they phrased it differently; a form pre-fills what can be read out of an uploaded document; the system points out that something does not add up.

The difference is in the brief. The goal is not to add an AI feature, but to walk the workflow and find the places where a person is still doing work a machine could have done.

It makes sense when

users cannot find what they need, and a search box has stopped being enough
people retype data that already exists in a document
onboarding a new user takes explaining rather than the product guiding them
you want AI in the product and want it to survive contact with real usage

What you get

AI-powered search and navigation over your own content
document processing — extracting data from PDFs and images
smart forms and assistance during data entry
an integration with Google Gemini, Vertex AI or Firebase AI
cost controls and defined behaviour when the model is unavailable

How it works

Find where it is worth it
We walk the workflow and pick the spots where assistance saves real time, instead of adding a feature for its own sake.
Try it on real data
A narrow scenario on your own documents or content shows what the quality will actually be before anything is committed to.
Build it into the product
The result lands inside the existing interface — in search, in the form, in the flow — not in a separate window.
Keep it running
Cost, quality and fallback behaviour get watched, because a model that changes underneath you is a normal operating condition.

Where you can see it

LingoSnap A sophisticated, AI-powered personal learning assistant designed to accelerate studying. It transforms content from images or PDFs into interactive materials like flashcards and quizzes using Google's Gemini API. Suited for language learners and students, it features multi-modal analysis, intelligent PDF parsing, and interactive tools with text-to-speech. Recut CV A stateless AI service that turns a candidate profile and a specific job advert into a resume tailored to the conventions of the target country. Its defining constraint is truthfulness: the model may select, reorder, rephrase and omit, but it may never introduce a number, employer, technology or date the source document does not contain — and that rule is enforced in code, not only in the prompt. It is in beta and not yet publicly available. ED Elektro A concept UX/UI redesign that turns a traditional electrical-materials wholesaler into an AI-first commerce platform. Instead of navigating a category tree, the customer starts with a plain-language sentence — 'I need to light up a warehouse' — and the interface answers with concrete products, live stock, and price per unit. The visual language is built from a manufacturer's catalogue sheet: calm, technically precise, one signal colour for what matters. Delivered as a full design system and a high-fidelity, clickable prototype.

Other services

SaaS products from idea to paying customersI build SaaS applications including the parts that usually take the longest: sign-up and authentication, subscriptions and pricing tiers via Stripe, and serving many customers from a single deployment. The same pieces are built into my own products.

Web applications and client portalsI build interfaces that combine product clarity, application logic and maintainable delivery.

Automated Angular 2+ refactoringI migrate older Angular 2+ codebases to modern Angular using AI-assisted tooling — quickly understanding the codebase, implementing changes with precision, and delivering results without tedious manual work.

UX/UI design and prototypingBefore a line of production code is written, the interface is validated on a prototype — wireframes, a clickable prototype and a design system that then holds up in the implementation.