We teach your team how to work with AI.

We learned to build software long before AI could, and now we build with it every day. Learn how to integrate the latest AI models in your workflow.

Three ways to work with us

If you don't know yet where AI fits, start with the AI scan. If your team already builds with AI, start with the coding assistants programme or the app check. If you pay for a tool or a website you would rather own, book a call.

Services
Service What you get
1Use AI in your daily work
AI scan A day in your office with the people who do the work. Then a report with three places where AI saves time, each with its cost, saving and risk. The report is yours, even if you stop there.1 day on site, report within a week
AI training for teams Half a day on your own documents: what you can paste where, which tool for what, where AI gets it wrong. You leave with a one-page AI policy.Half a day, up to 12 people
2Build with AI yourself, safely
AI coding assistants For development teams: Claude Code or Cursor set up on your repositories, with project rules, review habits and CI checks. We pair on your real backlog.2 days, or 4 remote half-days
Vibe-coded app check An engineer reviews the app you built with Lovable, Bolt, Cursor or another AI tool: login, database access, keys, where your data lives, backups. You get a report ranked by severity.3 days
Free checklist Eleven things to check yourself before your app goes live, with how and why. No email address needed.Online, right away
3Have us build it, in your name
Custom tools We build what the scan found, or replace one expensive tool. Code, accounts and data are in your name, documented so another developer can take over.Fixed price per project, maintenance monthly

Where the hours and the invoices go

Three sums you can redo with your own numbers, one for each way we work. Hours count at €48, the average labour cost in Belgium according to Eurostat (2024).

Copilot licences your team barely uses

Most offices already pay for Copilot or ChatGPT. Half a day on your own documents shows your team what to paste where, which tool fits which task and where it gets things wrong.

Half a day of training, up to 12 people, with a one-page AI policy.

The site-report app a colleague built with Lovable

An architect built a tool that turns a voice note into a site report. The hours are real, and so are the client names and site photos in it. Before the whole practice uses it, we check who can open what.

Vibe-coded app check: 3 days, then a report ranked by severity.

A job-planning app billed per technician

Your planners use its calendar and your technicians one job sheet. If our price for building and maintaining those two screens comes in under the subscription, we build them on your own accounts and you cancel it.

To build: about 10 to 18 days of work, then monthly maintenance.

All examples, with running costs and when it doesn't pay off

How it works

  1. A 30-minute call. On video or at your office, no slides. You tell us where work gets stuck.
  2. AI scan. One day with your team. You get the report within a week.
  3. You decide. Doing nothing is a fine outcome. If we build something, you know the price before we start.
  4. Build and hand over. Everything runs on your accounts, with documentation.

And if we're gone?

The code is in your repository, everything runs on your accounts, and how it works is written down. Another developer can take over without having to call us.

Where your data goes

Before we touch client data, we sign a data processing agreement.

Where your data goes Your team uses a tool on your own account in a European region. That tool sends only the text it needs to an AI model with a European endpoint, under a data processing agreement and with no training on your data. Nothing goes to Sinto Labs. EU Your team browser or app Your tool your account, EU region AI model EU endpoint data processing agreement no training on your data Sinto Labs no copy of your data access only while we work

We pick AI services with a European endpoint that don't train models on your data, and we tell you which ones up front. Health data and other special categories don't go to an AI model unless you explicitly decide they should.

How we handle your data

For those who already built something with AI

An app from Lovable, Bolt or Cursor often works straight away, but the AI assistant doesn't ask who else can reach your data. When a researcher scanned 1,645 Lovable apps in 2025, 170 of them had an open database.

Who leads Sinto Labs

Karolis Sinas

Karolis Sinas founded Sinto Labs in Ghent and leads it.

He built live-streaming and registration platforms before AI agents were usable, and builds with them now. He sets how we work and answers for what we deliver.

More about Sinto Labs