01 SIGNAL
02 MODEL
03 DECISION
AI & Data Studio · est. 2005 · Zagreb
Intelligence, engineered from data.
We build the models, pipelines and analytics that turn raw signal into decisions.
SCROLL TO ENTER ↓
TURNING COFFEE INTO CODE SINCE 2005
CHAPTER 01 — SIGNAL · RAW DATA
Most teams don't have a data problem. They have a decision problem — and the data is just where the answer is hiding.
21
YEARS TURNING COFFEE INTO CODE
2B+
ROWS PROCESSED DAILY
6
CORE AI & DATA PRACTICES
CHAPTER 02 — MODEL · THE ENGINE ROOM
Data that arrives fresh, clean and on time. Then we make it think.
01 / AI SOLUTIONS
Custom models & AI products
Bespoke models shaped around your domain — trained, evaluated and shipped into production.
02 / DATA SCIENCE
Numbers into narratives
Experiments, dashboards and the insight that moves a metric — an answer you can act on by Monday.
03 / DATA ENGINEERING
Pipelines & data platforms
Resilient ingestion, transformation and storage — the unglamorous layer everything else depends on.
04 / MACHINE LEARNING
Predictive & ML systems
Forecasting, scoring and recommendation engines that keep learning in production.
05 / GENERATIVE AI
RAG, copilots & agents
Grounded in your own knowledge — evaluated, and safe to put in front of customers.
06 / DATA STRATEGY
From spreadsheets to a data practice
Where to invest, what to measure, what to ignore — sequenced bets, not a wish list.
PHASE 01 · DISCOVERY · DATA AUDIT · SUCCESS METRIC
Frame
A fuzzy ambition becomes a sharp, measurable question — with an honest read on whether the data can answer it.
PHASE 02 · SPIKE · BASELINE MODEL · FEASIBILITY
Prototype
A fast, scrappy proof — enough to see the signal before anyone commits a production budget.
PHASE 03 · PIPELINES · EVALUATION · INTEGRATION
Build
Hardened into a real system: clean pipelines, evaluation you can trust, monitoring, and the interface people touch.
PHASE 04 · DOCS · ENABLEMENT · MONITORING
Handover
Runbooks and a working session with your team. The system keeps running — and improving — long after we've gone.
No black boxes.
2–4 WEEKS
Proof of concept
A focused sprint to prove the idea before you commit.
END TO END
Build & ship
We design, build and hand over a production system.
ON TAP
Embedded team
Our engineers inside yours — same standup, same goals.
CHAPTER 03 — DECISION · OUTCOMES
Decisions, not dashboards.
99.9%
PIPELINE UPTIME, MANAGED PLATFORMS
+32%
AVG. FORECAST ACCURACY LIFT
6 wks
MEDIAN KICKOFF → PRODUCTION
50M+
EVENTS PROCESSED EVERY DAY
100%
CODE & MODELS HANDED TO YOUR TEAM
0
BLACK BOXES SHIPPED, EVER
CASE · RETAIL DEMAND FORECASTING
−18%
food waste across 180 stores
Forecasting rebuilt on a modern pipeline for a regional grocery chain. +32% forecast accuracy.
CASE · LOGISTICS ANOMALY DETECTION
4h
earlier incident detection
Real-time detection over 50M freight events a day, owned and run by the client's team.
CASE · BANKING RAG COPILOT
70%
faster answers for support staff
An internal assistant grounded in 40,000 policy documents — evaluated and red-teamed.
[ BELIEF ]
“The best model is the one your team can run without us.”
LET'S BUILD
Have data?
Let's make it think.
info@blocksize.hrZagreb · Remote
CONTACT
Let's make it think.
Tell us the problem — the messier the better. We'll reply within one working day with the smallest thing worth building first.
EMAIL
info@blocksize.hrBASED IN
Zagreb · working remote
RESPONSE TIME
Within 1 working day
SINCE 2005 — COFFEE INTO CODE
Message sent.
Thanks — it's in our inbox. We'll reply within one working day.