AI Implementation

AI that ships to
production, not to slides

We build agents, RAG and automations that earn money and run in production — not ones that look pretty on slides. We join your team or take the whole project — from architecture to delivery.

AREAS

What we
actually ship

AGENTIC AI

Agentic AI

Autonomous agents in a loop doing real work — analysis, decisions, automation — in production, not in a demo.

agentsloopsproduction
RAG / KNOWLEDGE

RAG / knowledge

Retrieval over your documents and data; grounded answers, auditable, backed by evidence quotes.

retrievalgroundingcitations
AUTOMATION

Automation

LLM-powered document and process automation — with explainability: you know why the model made each call.

documentsprocessesexplainability
ROBOTICS

Robotics / embodied AI

Training and programming robots with agentic loops — embodied AI that does hard work, not a show.

agentic loopstrainingembodied
STACK

What we
build on

Python OpenAI Claude LangGraph RAG pgvector PostgreSQL Rails React Replicate ROS FastAPI
ENGAGEMENT MODELS

Two ways
we come in

AUGMENTATION

Team augmentation

We join your team for a concrete project — when hiring can't keep up; we plug into your stack and process and work as your people.

When hiring can't keep up and you're short on hands for a project
We plug into your stack, tools and process — no weeks of onboarding
We work as your people and report to your lead
You scale the team up or down depending on the project stage
PREFERRED
END-TO-END

End-to-end project delivery

We take the project from architecture to production and deliver the whole thing — you get the outcome, not a hiring problem.

We own delivery, the people and the quality — end to end
We run the project from architecture to production deployment
One accountable side — no managing vendors or recruitment
You get a working outcome, not a staffing problem to solve
A tight senior task force

An experienced task force,
not lone hands

Behind every delivery stands a tight task force of experienced freelance specialists who have shipped together for years — ML, backend, frontend, robotics, RAG, DevOps. Szymon runs it hands-on and matches the right seniors to each problem.

  • Battle-tested seniors — people vetted over years of real projects, not random contractors off a marketplace.
  • Matched to the problem — every task gets the right competencies, not the first free hands from a body shop.
  • Scaled to the stage — the team grows and shrinks with the project phase, so you do not overpay for slack.
  • We own the whole thing — we are accountable for end-to-end delivery, with no finger-pointing between vendors.

Ex-CTO of uPacjenta.pl (with an exit), co-founder of Diagnostyka Digital Hub, founder of RIOT — the first Polish Digital Product Studio — and former Head of Product at Golem Network.

ML / AI Backend Frontend Robotics RAG / Data DevOps
BILLING

Two models,
one decision

T&M

Time & Materials (billed per day)

You pay for work actually done — no more, no less; flexible scope. Best when the scope is still moving.

You pay for time actually worked — no more, no less
Flexible scope — re-prioritize at any moment
Best when scope still evolves and requirements emerge as you go
FIXED PRICE

Fixed price

Agreed scope, agreed price; budget certainty. Best when the scope is well defined.

Agreed scope and a price set upfront — no surprises
Full budget predictability from day one
Best when the scope is well defined and locked
WORK

See what
we've shipped

Global brand PoC · Agentic + RAG

Brand-message compliance control

Problem

The brand needs to know whether the messages appearing online are consistent with its PR strategy — scoring hundreds of publications by hand is slow and inconsistent.

Approach

An agent pulls publication links from media monitoring, extracts article content and metadata, then scores it against the brand's PR and brand theses (a brand playbook auto-built from the brand's documents as a RAG).

Outcome

Auditable verdicts: aligned / neutral / divergent — each backed by an evidence quote (proof of concept).

Python agent pipelineRAG (brand playbook)LLM scoringauditable output
DataPilot Investigation Studio Agentic + RAG

Digital investigation studio

Problem

Investigators need to surface evidence and connections from phone data — including hours of audio recordings to review.

Approach

A React/TS desktop app plus an audio transcription pipeline: cloud models via Replicate, speaker diarization, VAD chunking, and LLM correction of grammar and speaker labels.

Outcome

A working application: surfaces evidence and connections from phone data along with analysis-ready transcripts.

React 18 / Vite / TailwindPython agent backendtranscription pipeline (Replicate)
Medical distributor Production · LLM automation

Tender spreadsheet normalization

Problem

Every supplier sends a tender Excel file with a different header format — normalizing them by hand is slow.

Approach

The system detects tables in Excel, infers canonical tender headers with GPT, maps them structurally, and generates normalized, styled spreadsheets plus Markdown explainability logs (source→target mapping, confidence %, and the model's reasoning when confidence is low).

Outcome

Runs in production: normalized spreadsheets plus explainability logs for every decision.

Ruby on Rails 7SidekiqPostgreSQLPython + OpenAI
Security sector Robotics + agentic AI

Robotic security system

Problem

Physical security needs autonomous, always-on robotic monitoring — without a person constantly on site.

Approach

Agentic AI drives a robotic platform in a perception → decision → action loop, combining machine vision with autonomous decision-making.

Outcome

A built autonomous robotic security system running in a perception → decision → action loop.

robotics / embodied AIagentic control loopsperception / CV
OffBall Spatial AI · ML

Football spatial intelligence

Problem

Tracking-grade spatial analysis (pitch control, pressing, off-ball-run value, line-breaking) is out of reach for CEE and lower-tier clubs and agencies — they only have cheap on-ball event data.

Approach

Reconstructs the off-ball spatial layer from cheap event data: a physics-based spatial model (positions → pitch control etc., no training data needed) plus a light learned imputation step, calibrated and validated against paired event + tracking data.

Outcome

The make-or-break technical gate passed: a player-level Spearman > 0.7 target, achieving 0.83 (LOMO) on sealed paired matches under a pre-registered, adversarially-audited protocol.

Pythonphysics-based spatial modelevent-data MLvalidation protocol
Humanoid robotics tournament (Shenzhen) Robotics

A humanoid trained by agents

Problem

Training and programming a humanoid for an international robotics tournament — within a tight time window.

Approach

The entire robot training and programming was run by AI agents in a loop — the agents iterated on the robot's behavior right up to the start.

Outcome

3rd in Poland, top 32 of 75 teams from around the world — training fully run by agents in a loop.

agentic training loopsrobot controlembodied AI
PROCESS

From brief
to production

01

Brief

We understand the problem and the business context — what should happen and how we'll know it worked. You get a clear scope and goal.

02

Architecture

We design the solution: choice of models, data and integrations. You get the architecture and a plan before we start building.

03

Build / PoC

We build a working PoC or feature and iterate on real data. You get something that works and that you can actually touch.

04

Production & handoff

We ship to production and hand the project over with docs plus maintenance. You get a living system, not a demo.

A good fit if...

  • You have a real problem and data to work with
  • You want AI in production, not a demo on slides
  • You value auditability, explainability and team seniority
  • You need one person accountable for the whole thing
  • Your scope is clear or evolving — either billing model works

Not for you if...

  • You're after the cheapest body-leasing billed by the hour
  • You want buzzwords on slides with no real implementation
  • You have no data or access to build anything on
  • You expect magic without your own team's involvement
  • You're looking for a large, anonymous software house
NEXT STEP

Got a hard problem?
Let's talk.

A short call, a straight answer: whether it's doable, how, and in which model. One contact, one accountability.