NestLab · Founder-led AI studio

An EU AI Act scanner, before the 2 Dec 2027 obligations land.

An Annex III high-risk system has to be classified against duties that apply from 2 Dec 2027. The scanner does that today, in closed beta. The same founder-led studio has three more products in development and one open-source project with 320 stars and 120 forks.

How it works

From a description to a risk tier.

  1. The euaigap classify screen: system name, description, and a provisional high-risk result under Annex III point 4.
    01

    Describe the system

    Describe what it does, pick your role in the supply chain, and answer three screening questions.

    • Provider, deployer, importer or distributor — the role sets the tier
    • Free text, not a questionnaire: describe the system in your own words
    • Re-run any answer without starting the whole assessment over
  2. The euaigap result screen: high-risk AI system, Article 6(2) and Annex III point 4, with the deployer obligations listed.
    02

    Get the tier and its articles

    A risk tier with the articles it comes from and the reasoning attached. Rule-based, not a model guess.

    • Five tiers, from prohibited down to no risk
    • The articles the tier comes from, named
    • The reasoning written out, so it can be challenged
  3. The euaigap Conformity Assessment screen: the calculator inputs filled in for a high-risk system, next to the fine exposure it estimates - 100,000 to 150,000 euro under Article 99(4)(a) - and the QMS-based conformity assessment of Annex VI.
    03

    Size the exposure, then document it

    The penalty calculator prices the gap, and the documents follow from the same answers.

    • Fine exposure per Article 99, from the company size, sector and risk tier
    • Remediation cost next to the maximum exposure
    • Technical Report, Annex V Declaration, FRIA and Post-Market Monitoring Plan

Closed beta. Early access is by list.

Join the early-access list →
04
Flagship products
320
Stars on OpenStore
120
Forks of OpenStore
03
Infra regions — EU + North America
Principles

How we work.

Four anchors. Every product and every contract is tested against them.

01

Real products, not decks.

Every status on this site is one you can check: a public URL, a public repository, or a plain “in development”.

  • euaigap.co has been open to the public since June 2026
  • OpenStore is MIT-licensed, and the repository is public to clone
  • A product is marked “in development” until it is not
02

EU-first data residency.

Product runtimes on our own infrastructure — EU (Germany, France) for EU products, North America (Canada) for the North-America ones. Brevo for transactional email. Full posture on the security page.

  • Every subprocessor is named in public, with its processing location
  • TLS in transit, and every backup encrypted at rest
  • The hosting region is stated per product, not per marketing line
03

Stack you own.

Well-understood, widely used technology. No vendor lock-in. Our customers can migrate off without rewriting.

  • One codebase serves all four products, so no client forks it
  • A customer can migrate off without a rewrite
  • The open-source sibling runs on your own machine, not ours
04

AI scoped to measurable hours.

We add LLM calls where they save a documented hour, not for show. Every AI feature has a measurable ROI.

  • Every AI feature carries a documented hours-saved figure
  • Evaluation and cost limits are set before the build, not after
  • No LLM call anywhere that does not earn its place
What you get

What is in the box.

Every engagement ends in things you can point at, not a slide about them.

  • A written scope

    Success criteria, a fixed first milestone and an explicit out-of-scope list, signed off before any code ships.

  • Code you own

    Well-understood technology, no vendor lock-in. You can migrate off without a rewrite.

  • A production slice in four weeks

    One deployed feature path with auth, one core flow and one durable surface, end to end.

  • CI/CD from day one

    Wired on day one, not bolted on at the end.

  • Monitoring and a dashboard

    Uptime monitoring and host metrics. We hand off the operational dashboard, not just the code.

  • Encrypted backups

    Daily encrypted backups with a monthly restore drill.

  • EU data residency

    Product runtimes in the EU (Germany, France) by default; every subprocessor named in public.

  • A weekly demo, a monthly report

    A 30-minute call to demo what shipped, and one operational report: uptime, errors, cost.

Services

What we do, and who it is for.

Four ways to work with us. Each one names the buyer it is built for.

01

EU AI Act compliance

Preparation support for Regulation (EU) 2024/1689: risk classification, the Annex III duty set, and the documentation trail. Not legal advice.

Who buys it Compliance teams: CTO, DPO and procurement.

  • Annex III high-risk obligations apply from 2 Dec 2027
  • Named roles per service: provider, deployer, importer, distributor
02

AI workflow automation

Lead routing and follow-up, document intake, reporting, internal alerts and approvals — with a human-review path where a step needs judgement.

Who buys it Recruiters and HR teams processing CVs at volume.

  • An audit trail is part of the deliverable, not an add-on
  • A human-review step wherever a decision needs judgement
03

AI integration services

Retrieval-augmented search, assistants and extraction added to software you already run, with evaluation and cost limits.

Who buys it Brokerages and PropTech platforms.

  • Needs semantic listing search, ROI and CMA valuation
  • Evaluation and cost limits set before the build
04

Custom AI software development

A production-ready first slice in four weeks: auth, one core flow, deployment, monitoring and backups wired from day one.

Who buys it Teams that need one core flow in production first.

  • Auth, deployment, monitoring and backups wired from day one
  • Pricing written into the scope before kickoff

Not sure which of the four to start with? What to automate first is the checklist we run before scoping anything.

Planning for the EU AI Act? What applies to you is the checklist we run before scoping compliance work.

Flagship products

Two you can check today.

One has a public site, one has a public repository. The three below have neither yet, and say so.

The euaigap.co home screen.

EU AI Act Scanner

Closed beta · euaigap.co

For teams shipping an Annex III system.

Automated compliance assessment for the EU AI Act (Reg 2024/1689). Five-tier risk classification, gap analysis against Articles 9-15, four compliance documents generated.

  • 5-tier classification (prohibited → minimal)
  • 8-area scoring (risk mgmt, data gov, technical docs, record-keeping, transparency, human oversight, accuracy, robustness)
  • 4 doc types: Technical Report, Annex V Declaration, FRIA, Post-Market Monitoring Plan
  • Org role branching: provider, deployer, importer, distributor
Closed beta euaigap.co Join the early-access list →
The OpenStore home screen.

OpenStore

Open source · MIT

For teams who want to run the search themselves.

Open-source AI real-estate search. RAG over ChromaDB, 16 LLM providers, 9 languages. The OSS sibling of PropVector AI.

  • ChromaDB vector search with MMR reranking
  • 16 LLM providers behind one interface
  • 9 languages, pro-tuned AI prompts
  • Mortgage, ROI, CMA, TCO and multi-year price forecast
320 GitHub stars 120 forks MIT AleksNeStu/ai-real-estate-assistant

In development

No public URL yet. The product page for each says what exists today and who it is for.

The CheckCV dashboard: the intake board with the parsed CVs by state and the recent CV list.

CheckCV

In development

For recruiting teams.

Multi-format parsing, weighted rubric, candidate pipeline, white-label.

Read about CheckCV
The TuneCV home screen.

TuneCV

In development

For job seekers.

ATS scoring, AI suggestions, cover letters, interview prep, application tracker.

Read about TuneCV
The PropVector AI Agentic OS screen: the directory of available agents, each with its own description and example query.

PropVector AI

In development · OSS sibling live

For brokerages and PropTech platforms.

Vector search, MLS integration, 9 languages, ROI and CMA valuation.

Read about PropVector AI
Track record

Proof, not claims.

EU-sovereign, by default.

Product runtimes sit on our own EU infrastructure, not a shared platform.

  • TLS in transit, and every backup encrypted at rest
  • Every subprocessor listed publicly

Founder-led, no agency layer.

One operator runs the whole studio, and runs the products it ships.

  • One product in closed beta, three in development
  • One public open-source project
  • Engagements use the same stack as the products

Open by default, and priced before kickoff.

The posture and the price are both public before you sign anything.

  • Track record, security posture and subprocessor list are public
  • Pricing is written into the scope and does not move once signed
  • Six steps, every engagement, and the process is published too
Questions

Before you ask.

What does NestLab actually do?

NestLab is a founder-led AI studio, operating from Poland (EU) since January 2025. It builds and runs its own products — four flagship SaaS products plus one public open-source project — and takes on a limited number of client engagements using the same stack. Everything shipped here is also operated in production.

How does an engagement work?

Fixed scope and fixed price, written down before kickoff. The first milestone is scoped and priced before work begins, and the figure does not move once signed. Six steps, in order, every time.

Where does the data live?

Product runtimes run on dedicated infrastructure in the EU (Germany and France) and North America (Canada), with TLS in transit and backups encrypted at rest. This marketing site itself runs on Cloudflare Workers, a global anycast edge. Every subprocessor is listed publicly.

What technology do you build on?

Deliberately boring, well-understood technology, chosen so a system stays operable and portable years after launch. The specific components are shared with clients under the engagement, not published here. No vendor lock-in: your stack stays yours.

Who writes the code?

The person you email. No juniors, no handoffs. The work is done on boring, well-understood technology with no vendor lock-in, so the result stays maintainable by someone other than us.

What happens after launch?

You get the operational dashboard, not just the repository: uptime monitoring, daily encrypted backups with a monthly restore drill, and host metrics. Then weekly-to-monthly tune cycles against the metrics that matter to the business.

Start with a free 30-minute call.

We take on a limited number of contracts each year. Bring a process that is done by hand today, or a project you want to start — in half an hour we work out what it would take, and you leave with a written next step, whether that next step is with us or not.

  1. The process What is done by hand today, and by whom.
  2. The metric The one number that will say it worked.
  3. The deadline The date it has to work by.

Free. 30 minutes. A written next step, no obligation.