Skip to content

Why Gaspatchio?

Actuarial models live in three places today: spreadsheets that don't scale or version-control, vendor systems that lock the calculation behind a UI you don't own, and bespoke Python that takes an engineering team to keep performant. Gaspatchio is the Python option without the engineering-team requirement.

What you get

The model is Python; the execution is Rust. Write models in standard Python syntax, with standard libraries and a standard debugger. The execution runs on a Rust engine — portfolio-scale runs (100,000 policies × 30 years monthly) finish in seconds on the laptop you have, not overnight on a cluster.

Closed-form cumulative ops, plus a rollforward when you need state. For the calculations that fit the shape — survival probabilities, discount factors, AV with pre-computable charges — closed-form cumulative operations on list columns replace the recursive cell-graph. For the ones that don't — COI on net amount at risk, IUL floor/cap, GMDB ratchet, multi-state products — a rollforward() chain that reads like the within-period spec.

Auditable by default. Every output column traces back to inputs and intermediate steps. Schedules, assumption tables, curves, mortality tables, and compiled rollforwards each have an identity stamp — quarter-over-quarter drift is visible the moment it appears, not on page seventeen of an experience study.

Actuary-plus-LLM workflow that actually works. Python-native code is easy for AI coding tools to generate. Explicit calculations leave nothing for the LLM to hallucinate around. Error messages come back in actuarial language, and from your editor, gspio docs and gspio knowledge look up API methods and actuarial concepts directly — so the LLM working with you produces code you can review and sign off on, not a confidently-wrong intern.

No vendor lock-in. Open-source. The calculation is visible in the code. Your model travels with you between employers, audit firms, and regulators — no proprietary file format, no licensing renewal, no separate run-time you don't control.

When not to use it

  • You need a GUI-driven modelling environment for non-technical users.
  • Your team is committed to R as its stack.
  • You need a vendor with a ready library of jurisdiction-specific products today — Gaspatchio gives you the framework; you build the product layer on top.

Adjacent open-source projects

Other open-source actuarial work worth knowing about:

Where to go next

  • Tutorials — build a real model on the four-level ladder.
  • Concepts — understand how the framework thinks.
  • Modelling with AI — the actuary-plus-LLM workflow in detail.