01
Predictions from your data
We build models from the spreadsheets and databases you already have — accuracy that holds its own against black-box tools like Random Forest and XGBoost.
Mathematician · Business Developer · Tampere
Hi, I'm Reijo Jaakkola.
I help people build prediction models they can understand.
Business Developer at ExplainedAI, a commercialization project building interpretable machine learning solutions. Got something you'd like to predict? Let's talk.
Tampere UniversityBusiness Finland–fundedAward-winning research
Tampere, Finland
ExplainedAI
Where I work
Predictions you can explain
A Tampere University project, funded by Business Finland, that turns the data you already have into accurate predictions — each one backed by a reason a person can actually read. I'm the Business Developer.
Visit the ExplainedAI website ↗What we do
02
An explanation for every prediction
Each prediction comes with a short, human-readable reason you can inspect, audit, and defend — not a black box. Transparency is the whole point.
03
Cheap to run, compliant by design
Models light enough to train and run on a laptop — no GPUs — and transparent enough to meet EU AI Act and GDPR explainability requirements.
See it work
Real interpretable models — a handful of plain IF/THEN rules — matching or beating black-box methods on standard benchmarks.
Medical · Wisconsin Breast Cancer
IF Uniformity of cell size ≥ 3.5 OR Bare nuclei ≥ 3.5 AND cell size 1.5–3.5 OR Epithelial cell size ≥ 2.5 AND cell shape ≥ 3.5 THEN Malignant ELSE Benign
96.4%accuracy
matches Random Forest 97% · XGBoost 97%
Three rules a clinician can read — within a point of the black boxes.
Finance · German Credit
IF Loan ≥ 15.5 mo AND no history AND unemployed OR Loan ≥ 15.5 mo AND has checking account OR Has checking account AND no history AND no guarantor THEN Bad credit ELSE Good credit
60.7%F1 score
beats Random Forest 59.4% · XGBoost 58.9%
Catches more bad-credit cases than the black boxes — and tells you why.
Built on peer-reviewed research with collaborators at Tampere University, TU Wien, University of Helsinki & TU Dresden.
How a pilot works
Pilots are free — we're funded by Business Finland. You stay in control of your data the whole time.
1
We send you the library
A Python library that drops into your existing workflow — installed as a package, plugin, or API. No infrastructure changes.
2
You test it on your data
Train interpretable models on your own data, in-house. Nothing has to leave your systems — and it runs on a laptop, no GPU needed.
3
You get predictions + reasons
Accurate predictions, each with a short, human-readable rule behind it — benchmark it against your current models and see for yourself.
Common questions
The handful of things teams usually want cleared up before a pilot.
What does a pilot cost?
Nothing. ExplainedAI is funded by Business Finland, so pilots are free. You're not buying anything up front — you're checking whether interpretable models hold up on your own data before any commitment.
Does our data have to leave our systems?
No. You install the library and train models in-house, on your own machines. Nothing has to be uploaded or shared — your data stays exactly where it is, and you stay in control of it the whole time.
Will we lose accuracy by going interpretable?
On standard benchmarks, our models match or beat black-box methods like Random Forest and XGBoost. You don't have to take that on faith — during the pilot you benchmark them against your current models, on your own data, and see for yourself.
How is this different from explainability tools like SHAP?
Tools like SHAP try to approximate why a black box made a decision, after the fact. Our models are interpretable by construction — the prediction is a short set of plain IF/THEN rules, so there's nothing to estimate and nothing hidden.
What are the computing requirements?
Minimal. Models are light enough to train and run on an ordinary laptop — no GPUs or specialised hardware — and the library drops into an existing Python workflow as a package, plugin, or API. No infrastructure changes to get started.
Curious whether ExplainedAI fits your data? A short call is usually enough to tell.
Become a pilot partner →
Reijo Jaakkola
Who's Reijo
A mathematician who'd rather do useful work.
My path here ran through a decade of mathematics. It began with a master's thesis in mathematical logic that earned an Ernst Lindelöf Prize, then grew into doctoral research at Tampere University on logic and the theory of computation, with collaborators at TU Wien, the University of Helsinki, and TU Dresden. Across more than a dozen peer-reviewed papers, that work led me steadily toward a single idea — machine learning you can actually understand — and, eventually, to ExplainedAI.
Let's talk
Want to use your data for predictions?
Whether you want to try ExplainedAI's solution, need an outside opinion, have a speaking opportunity, or just want to get to know me — grab a time or drop me a line. I read every message.
Book a 20-minute call
The fastest way to find out whether ExplainedAI fits your data. No pitch — just a straight conversation. I usually reply within a day.
Email me to set up a time Opens your email app with a message ready to send