Tomasz Czarnecki
Tomasz Czarnecki Founding AI Product Lead · now building at Yaga Labs

I open black boxes.

I build AI people can check. Forecasts, agents and decision tools that people actually trust and use.

  • 1stProduct person in Heineken’s Global Analytics & AI Hub
  • 3AI systems shipped at Heineken
  • 36%ROI improvement across brands and touchpoints, per Heineken
  • 17+Markets shipped with Promo Advisor
  • 8 yearsWorking in China, client-facing in Mandarin
01RGM & AnalyticsHeineken · Global Analytics & AI Hub

Decision tools people actually use

Heineken spends about €7.9 million on marketing and sales a day. Someone decides where every one of those euros goes. The models were ready early. The people deciding still trusted their spreadsheets more.

Spent since midnight, your time €0 ~€2.9B a year ÷ 365 · FY2024 marketing & selling · Source

Case studies

Where to spend the marketing budget.

  • Bayesian allocation across markets, brands and channels.
  • More profitable spend with incremental gross profit reported in Mexico, Brazil and the Netherlands.
  • 36%ROI improvement across brands and touchpoints, per Heineken
  • 3Markets with gross-profit lift
  • BayesianResponse-curve model
Read the Allocation AI case study →
Allocation AI response-curve view: current spend sits past the point of peak ROI.

Then the measured lift decides next year’s budget.

02AboutNinety seconds, in person

First person in, three times.

The first hire at a Chinese architecture firm’s Polish office, set up in Poland from zero. One of two co-founders of MAD Architects’ computational design group in Beijing. The first product person in Heineken’s Global Analytics & AI Hub. Each time, the hard part wasn’t building the thing. It was getting people to use it.

Ninety seconds, in person.

Worked at and built for

Heineken MAD Architects AREP Macopedia Pearson Macmillan Education Raben
03OrchestrationAI orchestration, the ultimate skill

One person, a few agents.

I trained as an architect and wrote the parametric tools behind double-curved facades in Beijing. I spent eight years in China. Based in Guizhou as the client-facing lead, I brought in a ~70M RMB residential project in Mandarin, with a design team in Poland behind me. Now I build with AI agents, hands-on, every day. Product is where the three meet: I can shape what people see, earn the trust of the people who will use it, and build the first version myself.

AI orchestration is the ultimate skill. Hermes is my orchestrator. It lives on a Mac Mini, works there and is always on, 24 hours a day. Everything I run goes through three agents. Each one has its own skills, its own purpose, its own SOUL.md and its own memory. All of them are model-agnostic.

Tomasz’s avatar: a cartoon man with short grey-brown hair in a black hoodie on an orange backgroundTomaszOne person, AI orchestrator
  • ShapeArchitecture, 3D, parametric
  • BuildAI, agents, open source
  • BridgeMandarin, China ↔ Europe
  • Product3 AI systems shipped at Heineken
Hermes Agent logo
Orchestrator

Hermes

Lives and works on a Mac Mini. It never switches off.

  • Mac Mini
  • Always on, 24 h
  • Model-agnostic
  • Lily’s avatar: a cartoon woman with a blonde bob in a black hoodie on a pink backgroundLilyPersonal assistantCreates music and video, and orchestrates my AI brain.
  • Yaga’s avatar: a cartoon old woman with red curls and round glasses in a black hoodie on a green backgroundYagaBuilderSpecialises in building things and writing code.
  • Anna’s avatar: a cartoon woman with a dark bob in a green jumper on a blue backgroundAnnaMy wife’s assistantLives alongside Lily and Yaga.

Each agent has its own

  • SkillsWhat it can do.
  • PurposeWhat it is for.
  • SOUL.mdWho it is and how it talks.
  • MemoryWhat it remembers.
04AI ProjectsBuilt with AI agents

Selected projects.

Side projects, built with AI agents. Some are live and some are still in progress.

A wedding party with lanterns on the cobbled market square at night The fortress town from above at night, ringed by water The castle and square under snow at dusk The church on a snowy square, with horse-drawn carriages The market square in the rain, seen from an upper window

01Historical reconstruction · Live

Küstrin 1913

A 3D reconstruction of the fortress town, destroyed in 1945 and never rebuilt. Agents rebuilt it from old photographs, maps and records.

The palace gardens from high above: lakes, islands and courtyards The European-style palace with fountains and formal gardens A pavilion island on the lake, with boats and willows

02Historical reconstruction · In progress

Old Summer Palace 圆明园

A 3D reconstruction of the emperors’ garden palace in Beijing, burned in 1860.

Zoja in black leather and a spiked crown, holding a microphone on a spiked black throne Zoja in a black robe, singing before an open gothic gate in a misty forest Zoja in white face paint and a spiked crown, reaching out in a foggy forest

03AI music · Live

Zoja Music

Slavic folk metal sung in Interslavic, a language most Slavs can follow without learning it. The singer, music and videos are made with AI.

卓娅 reclining on a plum velvet chaise with a bouquet Close-up portrait of 卓娅 holding a pale dahlia 卓娅 in a pale blue dress on a blue tufted sofa

04AI persona · Live

卓娅

My AI influencer on Xiaohongshu: a face, a style and thousands of real followers.

QR code for 卓娅 on Xiaohongshu

Scan with Xiaohongshu to find her.
Rednote ID 42044731384

More projects

05ChinaAlgorithmic Design

Back in China, the fitness function was the eval.

Eight years working in China, back and forth ever since. I wrote the scripts behind a facade and an opera house interior at MAD, and competition wins at AREP. A decade before “AI agent” was a phrase, I was building evolutionary solvers that bred building geometry from a fitness function.

Selected buildings

Images courtesy of MAD Architects and AREP

Live model The Xinhee facade, evolving An evolutionary solver breeds the membrane geometry, scoring every generation on cost and buildability.
  • MAD Architects team photo in front of the Beijing office
    MAD family
  • Tomasz laughing with teammates around a monitor at MAD
    BaiZiWan team
  • Tomasz with five MAD colleagues after work
    After work
  • Tomasz with his Gangxia North hub team in the office
    Gangxia team
  • Tomasz lying on the model table, adjusting a massing model
    On the model

Optimal was easy. Buildable was the job.

06Yaga LabsBusiness case

Where I build now.

AI-first software house

Yaga Labs

An AI-native team for larger and regulated companies.

  • Agentic workflows and AI software, designed, built and run.
  • Models read and summarise, your rules decide, and a person signs off where it matters.
Visit Yaga Labs ↗
Yaga Labs logo: a birch broom with a red binding forms the slash in a </> code tag
07Contact

Got a black box nobody uses?

Tell me what it predicts and who ignores it. I’ll tell you what I’d build first.

tomasz@czarnecki.ai