An AI MVP in 14 days

A working AI product in two weeks. At a fixed price.

Have an idea for a product or a process to automate? In 14 days you get a working AI application in your users' hands, not a slide deck and an estimate.

Maciej Krupowies

Run by
Maciej Krupowies · Top 1% on Upwork · LinkedIn

Experience from projects for

  • GSK
  • TikTok
  • Google
  • Adobe
  • Mozilla
  • LPP
  • Voicelab.ai
  • Tunity

Calculator

How many hours will your team get back?

Pick a process and enter your numbers. You'll see how much time and money AI saves. Your people spend the hours they get back on customers, sales and growth instead of retyping.

0hours back every month
0of a full-time job freed up
0saved per year
-MVP pays back in

Payback based on an MVP from PLN 34,000 net. An estimate from typical values. On a call we'll run it on your own data.

What you get

  • A web or mobile app with AI: chat, agent, document analysis or voice
  • Cloud deployment with login, payments and an admin panel
  • Code in your repository. You own it
  • A demo every two days, so progress is always visible
  • Handover and a short training for your team

Technology

OpenAI, Claude and open-source modelsRAG and document searchAgents and automationVoice agentsFastAPI, React, ExpoAWS, Terraform, Docker

Packages

Prototype in 5 days

One key AI feature

from PLN 14,000 net

  • Runs on real data
  • Ready for user testing
  • A recommendation: build further or not

Post-launch care

Growth and maintenance

from PLN 4,000 net / month

  • New features at a steady pace
  • Monitoring, fixes and model costs under control
  • No multi-year contract

How we work

  1. 1

    Day 0: call and scope

    A 30-minute call, then a written scope, timeline and fixed price.

  2. 2

    Days 1–2: design

    Wireframes, architecture and data plan. You approve before we build.

  3. 3

    Days 3–12: build

    A demo every two days on a working version, not on slides.

  4. 4

    Days 13–14: launch

    Production deployment, code handover and team training.

Security and compliance

Cybersecurity from day one, not after the audit

Hands-on cybersecurity experience and projects in regulated industries: finance, healthcare, compliance. Systems are built to the requirements of ISO 27001, SOC 2, HIPAA and GDPR, so they pass audits without rework.

  • ISO 27001
  • SOC 2
  • HIPAA
  • GDPR
  • EU AI Act

Data under control

Encryption at rest and in transit, EU data residency when you need it, retention and deletion policies.

Access and audit trail

Roles and permissions, event logging, access reviews and logs ready for your auditor.

AI without leaks

Enterprise model tiers or models in your own cloud, no training on your data, personal data masking.

Audit-ready documentation

Architecture, data flows, risk register and the procedures auditors ask for.

Certification is carried out by an accredited body or auditor. We prepare the system and documentation so they pass.

Selected projects

Client project · Financial services

Credit underwriting copilot

A copilot that reads loan documents, computes the ratios and drafts the credit memo for the analyst.

RAGpgvectorOpenAIClaude

Client project · Manufacturing · Poland

From client brief to quote in a minute

For a cosmetics manufacturer: AI reads client briefs and labels, extracts product parameters with confidence scores and produces a ready quote PDF.

LLMOCRRAGPDF

Client project · Financial compliance

AML and KYC checks with an AI memo

Client intake, sanctions screening, a risk score and a draft compliance memo written by AI.

ClaudeAML/KYCAngularFastAPI

Client project · Wearables

Voice AI in earbuds

A voice AI assistant on wireless earbuds with near-zero latency.

STT/TTSEdge AIiOSAndroid

Client project · Corporate learning

An AI coach grounded in the curriculum

A platform where AI answers employees using only the company's approved training material.

RAGOpenAIClaudeWebSocket

Client project · Document automation

Document classification and extraction

A system that recognises each document type and extracts its data with no manual retyping.

ClaudeAWS BedrockLambda

Frequently asked questions

What if the scope changes along the way?

We agree the scope before starting. Changes mid-way are quoted separately or moved to the next phase, so the deadline holds.

Who owns the code?

You do, entirely, from handover.

What if 14 days is not enough?

I say so on the first call and propose phases. I don't take projects that can't be delivered in that time.

What does running the AI cost?

It depends on your user numbers. Before we start you get an estimate of monthly model and infrastructure costs.

Contact

Let's talk specifics

Tell me in two sentences what you need. I reply within one working day.

or email office@bracketslabs.com

Maciej Krupowies
Maciej Krupowies
Brackets Labs

I use the details from this form only to reply to your message. Details in the privacy policy.