From Excel Tariff Calculators to Python – Agentic Edition

Einleitung/Dauer

Excel-based tariff calculators remain a cornerstone of actuarial work in life insurance, and in early 2026 we showed in our first EAA web session 'From Excel Tariff Calculators to LLM-Powered Python Code', how Large Language Models (LLMs) can help to port them into structured Python code. The feedback was encouraging, and the field has moved forward quickly since then. Tools have matured, reasoning quality has improved, and, most importantly, the working practice has shifted from a single actuary prompting an LLM into an orchestrated collaboration between actuary and AI agents.

This second edition takes that step forward. Instead of a craftsmanship-style workflow, we present an industrial, agent-based pipeline for calculator porting: structured extraction from Excel and VBA, preparation of actuarial context, LLM-driven code and test generation, and built-in quality assurance, all orchestrated so that the actuary keeps the lead and the LLM handles bounded subtasks autonomously. The pipeline itself has been built inside our DAV working-group context on AI-assisted calculation-engine development.

Participants will see the framework in action on a real-life insurance tariff calculator, from raw Excel input to a ported Python module with a matching test harness. We will not only show the happy path, but also the places where quality control hooks in, how disagreements between LLM-generated output and reference values are resolved, and how the actuary stays firmly accountable for what the pipeline produces.

This web session closes with a hands-on exercise in which each participant uses the pipeline to build a test comparison layer for a given piece of ported code. Because the pipeline is already in place, the LLM focuses narrowly on generating the test framework and the reference calculations needed to validate the port. A well-scoped task that gives participants an immediate, tangible experience of what agentic actuarial work feels like in practice.

Preliminary Programme
Friday, 27 November 2026
09:30 – 10:30  From craftsmanship to industry — what we have learned since February 2026
10:30 – 10:45  Break
10:45 – 12:15  The agentic pipeline: architecture, roles, live end-to-end demo on a life-insurance tariff calculator
12:15 – 12:45  Break
12:45 – 14:00  Hands-on: building a test comparison layer with LLM support
14:00 – 14:15  Wrap-up, open questions, outlook

All the above times are given in CET (Central European Time).

Vorgehensweise und Ziele

The purpose of this web session is to make the transition from exploratory LLM use to industrial-grade AI-assisted calculation-engine development concrete and tangible. Many actuarial teams have now experimented with LLMs on individual tasks, but struggle to turn these experiments into a repeatable, trustworthy process. This session addresses that gap directly.

Participants will see a working pipeline that has been built inside an actuarial working group, understand the design principles behind it, and take part in one of its steps themselves. The emphasis is less on promoting a specific tool stack and more on the underlying method: what stays with the actuary, what can safely be delegated to agents, and how quality assurance holds the whole thing together.

The web session is designed to leave participants with a clear mental model of what an industrial agentic actuarial workflow looks like including a concrete starting point for building similar workflows in their own organisations.

Teilnehmer

This web session is aimed at actuaries, product developers, and modelling teams in the insurance industry, particularly in life insurance, who want to go beyond individual LLM prompting and understand how AI can be embedded systematically into calculation-engine development. It will also be valuable for teams running portfolio migrations or consolidation projects where manual Excel-to-code conversion has become a bottleneck.

Basic Python knowledge is helpful for following along with the live demonstration and the hands-on part. Prior LLM experience is not required. All concepts, tools, and workflow steps will be introduced step by step, and the hands-on exercise is scoped so that every participant can complete it within the allocated time.

Technical Requirements
Please check with your IT department if your firewall and computer settings support web session participation (the programme Zoom will be used for this online training). Please also make sure to join the web session with a stable internet connection.

Dozierende

Bartlomiej Maciaga
Bartlomiej Maciaga is a qualified actuary with over 20 years of professional experience in life insurance, specialising in product development, IT migrations, and actuarial process automation. He chairs the DAV sub-working group on White Box AI in portfolio migration and leads the sub-sub-working group on AI-assisted calculation-engine development. His current focus is on building reproducible, industrial-grade pipelines in which LLMs support actuarial work under clear methodological guardrails.

Dr. Arno Rasch
Arno Rasch is the CEO of vtmw AG, a consulting company with longstanding experience exclusively for insurance companies, with particular expertise in portfolio migrations. Under his responsibility, the firm increasingly applies artificial intelligence and automation to support migration projects and other areas of consulting. He holds a PhD in Computer Science from RWTH Aachen University, focuses on automation and AI applications in the insurance sector, and is co-editor of the yearbook Insurance & Innovation.

Sprache/Kurztitel

The language of the web session will be English.

CPD Credits
For this web session, the following CPD credits are available under the CPD scheme of the relevant national actuarial association:

• Austria: 4 points
• Belgium:  4 points
• Bulgaria:  6 points
• Croatia: individual accreditation
• Czechia:  4 hours
• Denmark 4 credits
• Estonia:  4 hours
• Finland: 4 points
• France:  24 points
• Germany:  4 hours
• Greece: 5 points
• Hungary:  4 hours
• Iceland: 4 credits
• Ireland: 4 hours
• Italy: individual accreditation
• Latvia: 4 hours
• Lithuania: 4 hours
• Netherlands:  approx. 4 points (individual accreditation)
• Norway: 4 points
• Poland: 4 hours
• Portugal: 4 hours
• Serbia: 4 hours
• Slovakia:  individual accreditation
• Slovenia:  individual accreditation
• Spain: CAC: 4 hours, IAE: 4 hours
• Switzerland:  individual accreditation
• USA: SOA (Section B): up to 4.8 hours

No responsibility is taken for the accuracy of this information.