PredicTri
Next-Gen Triangle Modeling with AI/ML
Finally, an AI-Powered Reserving Tool You Can Trust
Built by actuaries, for actuaries, to bring clarity and confidence to your results.
๐ฅ Reliable Results
Modern Bayesian ML provides a trustworthy second opinion on reserves.
๐ฅ Full Transparency
No black boxes. Get clear diagnostics and confidence intervals.
๐ฅ Model Validation
PredicTri tests multiple models and automatically selects the best one.
โก And More...
๐ก Why PredicTri?
PredicTri is a cloud-based actuarial tool that brings Bayesian machine learning to triangle reserving.
In simple terms: we help insurers predict future claims costs more accurately and transparently, helping manage risk and allocate capital more effectively.
Designed for actuaries, PredicTri adds a second layer of insight and validation to your existing workflow, improving accuracy and explainability where it matters most.
Accurate Results, Backed by Bayesian Inference
PredicTri delivers reliable results using advanced Bayesian modeling.
Model Validation & Selection
PredicTri evaluates multiple models and automatically selects the best-performing one based on predictive power.
Second Opinion / Peer Review
Use PredicTri as a second-opinion tool to review traditional results.
Transparent Outputs
Transparent outputs with confidence intervals and model diagnostics โ no black box models.
Regulatory Ready
Designed for accuracy, explainability, and regulatory support with full documentation.
Small Portfolio Compatible
Compatible with small or volatile portfolios where traditional methods struggle.
Workflow Enhancement
Enhances your existing workflow without replacing your models or professional judgment.
Joint Modeling
Joint modeling of paid and incurred triangles for comprehensive analysis and improved accuracy.
๐ What You'll Find Here:
One-Pager
A concise overview of what PredicTri does and why it matters for actuaries. โก๏ธ Download the One-Pager
Case Study
See how PredicTri helped identify hidden patterns in a real-world portfolio. โก๏ธ Read the Case Study
Input & Output Instructions
Simple guidance on how to upload your triangle and understand the results. โก๏ธ View Input/Output Instructions
Video Walkthrough
A quick demo showing how to use PredicTri. โก๏ธ Watch on YouTube
Content Library
Comprehensive collection of articles, case studies, and insights on AI/ML in actuarial practice. โก๏ธ Browse Content Library
๐ Want to Try PredicTri?
Get in touch with our team to request a trial and see how PredicTri can improve your actuarial workflow.
๐ฌ Join the Discussion
We're building a community of actuaries shaping the future of reserving. Connect with fellow professionals, share insights, and discuss the latest developments in AI/ML applications for actuarial science.
Join LinkedIn GroupAbout PredicTri
PredicTri is a software tool for non-life insurance reserving. It takes the same input data actuaries already use, paid and incurred claims triangles, and applies advanced Bayesian machine learning. The approach captures uncertainty, structural changes, and correlations that standard deterministic or stochastic triangle methods often miss. PredicTri provides actuaries with an explainable second opinion that can be directly compared with existing calculations, helping them validate assumptions and clearly explain results.
How It Works:
Modelling Flow

- PredicTri starts with the input data: paid and/or incurred claims triangles together with premium or other exposure measures.
- Next, PredicTri runs the data through multiple Bayesian models.
- Each model is tested for how well it predicts unseen parts of the data, using a cross-validation method.
- The system selects the model with the strongest predictive power.
- Finally, the results are presented through clear reports and analytics that explain both the numbers and the drivers behind them.
Business Flow
PredicTri links technical reserving with business outcomes. You begin with your usual risk segmentation and exposure build-up, then model aggregate losses with PredicTri. This produces transparent reserves and capital requirements. Once the capital picture is clear, you can overlay pricing to evaluate profitability and strategic outcomes.

- Define risk groups or exposure categories; some initial modeling can already be applied at this stage.
- Build portfolio aggregates.
- Model aggregate losses with PredicTri.
- Assess reserves and capital requirements.
- Overlay pricing to understand business outcomes.
Key Benefits:
- Enhanced accuracy through Bayesian machine learning
- Full transparency with confidence intervals and diagnostics
- Regulatory compliance and explainable results
- Seamless integration with existing workflows
- Designed specifically for actuarial professionals