Kodulabor

Applied AI research. Measured outcomes. Published results.

Kodulabor is an independent research lab that runs controlled experiments on real business workflows. We select a small number of projects each quarter, apply AI under rigorous conditions, measure everything, and publish the results — including what didn't work.

Latest case study

Revalia Homes

A multilingual real estate platform with intelligent content management. Built, tested, and deployed in a single day with approximately 5–6 hours of active human effort and ~€50 in API costs.

789

Properties indexed

Languages supported

5–6 hours

Human effort

~€50

Total cost

27x

Acceleration

Read the full case study →

How it works

Each project follows the same protocol. Define a real workflow. Apply AI under controlled conditions. Measure time, cost, quality, and failure modes. Publish the full assessment — numbers, gotchas, and an honest verdict on whether it actually helped. No hype. No sales pitch. Just evidence.

Problems we investigate

We look for workflows where AI might produce a measurable improvement — and where the results are worth documenting for others.

—Repetitive knowledge work that follows recognizable patterns

—Content operations — translations, summaries, formatting — that repeat every cycle

—Manual reporting that consumes hours but produces predictable outputs

—Website and application builds where speed-to-production matters

—Operational bottlenecks in small teams where automation could shift capacity

—Feasibility questions — whether AI actually helps a specific workflow, tested with real data

Browse case studies See the framework Propose a project