CAPITAL
AT RISK
Capital at Risk - Applied Research

HAL

A financial plan you can question.

Matthew Lynch· 14 Oct 2026

Abstract

Ask a retirement question in plain English. HAL starts with what you want to achieve, asks for the information needed to test your choices and explains the result. A governed methodology sets what can be concluded, Engine B does every calculation, and the decisions stay with you. HAL is a research demonstration of Interactive Financial Planning: conversation on the surface, governed calculation underneath.

Why this began

The research began with my work in retirement cashflow planning. I wanted a more rigorous approach to selecting, evidencing and challenging the assumptions on which financial decisions depend.

That became the Capital at Risk methodology: a comprehensive set of rules and technical guidance, supported by an Execution Sheet to make the process usable in practice.

But applying a sound methodology was only part of the challenge. The mechanics of cashflow modelling could still interrupt the client conversation.

Take a simple question: “What happens if I retire a year later?”

In conventional cashflow software, answering it may require changing inputs, creating another scenario, recalculating and comparing the results, while the client waits.

I developed Engine B to execute the methodology consistently, and HAL to make it accessible through conversation. The question can be asked naturally, the calculation runs and the chart updates.

The adviser can concentrate on exploring the financial implications with the client rather than operating the software.

The FCA’s cashflow modelling findings (20 March 2024) also emphasised input accuracy, justified assumptions, uncertainty and the review of outputs.

Governance before calculation

A cashflow projection is only as good as the assumptions fed into it. The methodology settles those assumptions, and the evidence needed before any conclusion is allowed, before a model is run. It is four documents:

Rules
What must be true for a conclusion to stand
Register
The numbers: growth, stress, charges, inflation
Execution Sheet
How a practitioner runs a case
Manual
Why each rule exists, and the evidence behind it

The FCA’s October 2026 findings on AI governance highlighted inadequate documentation, weak audit trails and insufficient challenge of vendor assurances. HAL’s architecture separates conversational AI from the methodology and calculation engine, retaining the inputs, assumptions and evidence behind each result. This provides inspectability, although it does not, by itself, establish compliance with the FCA’s wider expectations for AI governance.

The methodology distinguishes financial planning from formal grading. A Current projection shows where the client stands today and is not graded. Clients and advisers can develop and explore financial plans without a formal grade. An Advised plan, where an adviser recommends a strategy, is formally graded against the rules.

The methodology sets projections in today’s money, running to age 99, with investment stresses calibrated to the agreed risk level and their financial consequences assessed against the client’s objectives.

Selected pages from the Manual
1 / 9
Read the technical methodologyControlled access
Technical Methodology & Evidence Manual, version 1.0. Opening pages, Chapter 1 and the start of Chapter 4. Select a page to read it.

Making the methodology executable

A substantial technical manual cannot sit beside every adviser in every meeting. Engine B applies the rules in code.

The same inputs on the same engine version give the same answer. Every result carries the inputs, assumptions and controls that produced it.

When a conclusion is not supported, Engine B says why, and tells three cases apart: evidence is missing, no strategy has been set, or the circumstances do not justify a grade. It applies governed assumptions. It never invents client facts.

Engine B was tested against independently constructed oracles.

  1. Foundation Capital at Risk Methodology Sets the evidence requirements, assumptions, planning rules and decision controls
  2. Execution Engine B Runs governed calculations and stress tests, identifies considerations and records evidence
  3. Application HAL AI conversation and planning intelligence (M-PI). Selects relevant considerations, explores alternatives and explains results
  4. Judgement Client / Adviser Supplies information, evaluates choices and retains the decision
Figure 1. The methodology sets the rules. Engine B performs the calculations and identifies considerations. HAL combines conversational AI with M-PI for governed exploration. The client and adviser retain judgement and decisions.

Making it conversational

HAL starts by establishing what you’re trying to achieve, rather than taking you straight into a financial fact-find. It takes a question as asked, works out what is changing, asks for the information it needs and has Engine B run it. Then the chart updates.

The language model never does the arithmetic. It handles the conversation; Engine B is the calculation authority. The conversational AI uses a replaceable model API. The methodology, Engine B calculations and M-PI’s consideration-selection rules remain separate from the language model.

Any reply backed by a calculation can be opened in the Calculation Inspector, which shows the inputs, assumptions and controls behind the number.

Recording of HAL answering a demo client from the live engine. Real output only.

Two ways to use HAL

On your own. Ask about your financial plan in plain English. HAL builds the picture through conversation, without requiring you to learn financial-planning software or complete a form first.

With an adviser. Describe the case and ask the question. Engine B applies the configured methodology and assumptions automatically. Ask another question, test an alternative and compare the results without rebuilding scenarios or cloning cases. The original plan is preserved, and the calculations, assumptions and evidence can be inspected.

The adviser retains responsibility for professional judgement and any recommendation.

The alpha uses Capital at Risk’s methodology and assumptions. Firm-specific configuration is a future capability.

Does the gap matter here?

Not every missing piece of information prevents a financial conclusion. Where the methodology permits alternative readings, Engine B can test whether the uncertainty affects the decision.

If the relevant outcomes remain unchanged across those readings, the calculation can proceed where the rules permit, with the uncertainty recorded. If the outcomes differ, further information is required before a supported conclusion can be reached.

Some information is fundamental to the calculation itself. Without an established expenditure requirement, for example, retirement affordability cannot be properly assessed.

What does the methodology permit?

  1. Sufficient Engine B calculates and returns a supported conclusion
  2. Missing, but immaterial Proceed where the methodology permits, recording the uncertainty
  3. Missing and material HOLD. HAL asks the next necessary question
  4. No strategy, or no justified grade HAL explains the limitation. No conclusion is manufactured
Figure 2. Material gaps can prevent a supported conclusion. Immaterial gaps may be recorded without changing the decision, where the methodology permits.

Try HAL

Start with one of the three prepared households: Helen Ward, Ravi and Maya Shah, or Claire and David Bennett. Their details are already recorded, so you can ask about the plan straight away.

Ask what you would ask an adviser. Are we on track? What if we retire later, spend less or give more away? How long would I need to keep working? HAL answers in a sentence and the plan shows what changed. Where an answer rests on a calculation, the cog beneath it opens the Calculation Inspector with the workings.

Want to test it harder? Start as a new client and describe a fictional household of your own. HAL asks for what it needs, takes you through the attitude-to-risk questionnaire and builds the plan from your answers. It takes longer. It tests whether HAL can build and analyse a plan from information it has never seen before.

HAL is a research demonstration, not financial advice. You will be asked to confirm this before you begin.

What is established, and what is not

Engine B is tested in two ways. For determinism, independent runs on the same inputs and engine version must reproduce identical outputs and trace hashes. For conformance, its conclusions are checked against expected results constructed independently of the engine.

HAL and Engine B are tested together, end to end, before release.

Not claimed

  • That any output is financial advice or a suitability assessment.
  • That the methodology has been independently reviewed.
  • That independent implementations have converged.
  • That the economic assumptions are calibrated against external benchmarks.
  • That HAL is a financial adviser, or knows what is best for a household.
  • That comparing options is a personal recommendation.
  • That HAL meets the FCA’s expectations for a firm’s governance of AI.
HAL is a research demonstration. Nothing on this page is financial advice or a personal recommendation, and no output should be relied on to make a financial decision. All households shown are synthetic or fictional.
© 2026 Capital at Risk. All rights reserved.