
1. Technology overview
Probabl is a Paris-based company (probabl SAS, SIREN 979 093 523) that describes itself as the tabular AI company and stewards the scikit-learn project. It addresses the difficulty of building reliable predictive models on enterprise tabular data.
Its Skore platform tracks experiments, compares runs, reports uncertainty and run cost, and lets AI agents suggest next steps within configurable autonomy. Read-only connectors reach Snowflake, Postgres, Databricks and BigQuery, with an audit log for governance. An MCP version and a scikit-learn catalogue complete the offer.
2. Highlighted features & specifications
- Assist: agent suggestions that flag pitfalls and propose next steps.
- Delegate: configurable autonomy with an end-to-end run mode and token limits.
- Experiments: automatic run tracking and side-by-side comparison.
- Measure: metric selection, uncertainty reporting and run cost tracking.
- Secure: read-only data connectors and vault-stored credentials.
- Govern: timestamped agent journal and audit log.
3. Potential application
A Vietnamese bank data team can track and compare credit-scoring models in one workspace. A retailer can build demand-forecast models on warehouse data in Snowflake or BigQuery. A manufacturer can audit predictive-maintenance experiments.
4. Benefits
Automatic experiment tracking and side-by-side comparison make model results reproducible. Uncertainty and cost reporting support decisions on which model to deploy. Read-only connectors and workspace-scoped credentials limit data-access risk. Agent actions are logged in a timestamped journal for audit.
5. Market trends & opportunities
Most enterprise data is tabular, and companies seek predictive models that are auditable and cheaper than generative AI. Demand for governed AI agents on structured data is rising. Vietnamese banks, retailers and manufacturers hold large tabular datasets for scoring and forecasting.
6. Other information
probabl SAS, SIREN 979 093 523, Paris. Seed round of EUR 13M announced 2025-10. Collaboration with Intel on scikit-learn announced 2026-06.
One platform connecting startups, corporations, research institutions, regions and investors











Playbook & Handbook · February 6, 2026
Winning Accelerator Playbook 2026 by BambuUP
RETAIL E-COMMERCE: AI AND REAL-TIME DATA ARE TRANSFORMING HOW RETAIL BUSINESSES OPERATE








































