# Personal Finance App pitch
> An open-source financial twin emerging from personal use and agent-first development
- HTML version: https://robbiepalmer.me/projects/personal-finance-app/deck
- Project: Personal Finance App (https://robbiepalmer.me/projects/personal-finance-app.md)
- Slides: 15
## Slide 1
# A financial twin for the whole household
Asset Tracker keeps assets, debts, ownership, currencies, plans and outcomes in one inspectable model.
It began with a jointly owned home, unequal deposits, a mortgage approaching renewal, and investments spread across
GBP and USD.
## Slide 2
## My spreadsheet is doing four jobs
01LedgerWhat existed, changed or closed?
02Ownership graphWho owns the asset, and what debt is linked to it?
03Planning modelWhat changes if rates, work or contributions change?
04ReconciliationDid reality follow the plan?
Every exception becomes another tab, formula and assumption to maintain.
## Slide 3
## The prototype already helps me
RecordFinancial historyBalances, holdings, closures and imports
FlowsMoney movingTransfers, capital flows and currencies
HistoryPosition over timeNet worth, expenditure and FI progress
DecideWhat is possiblePortfolio projections and liquid runway
It is a client-side prototype in active personal use. Product-market fit remains unproved.
## Slide 4
## The model should match the household
```mermaid
flowchart TB
Household["People · ownership
responsibility"] --> State["Assets · debts · accounts
currencies"]
State --> History["Balances · holdings
transfers · income · spending"]
History --> Decision["Runway · mortgage · work
investing"]
Plan["Plans · assumptions"] --> Decision
Decision --> Outcome["What happened next"]
Outcome --> History
```
The plan and the actual financial history should update the same source of truth.
## Slide 5
## Three products hide inside "personal finance"
01Account appWhere is the money now?
02Planning toolDoes this scenario work?
03Financial twinWhat changed, and what happened next?
The twin keeps household state and decisions on one timeline.
## Slide 6
## The category is occupied
MonzoDistribution and live financial data
ProjectionLabPlanning, tax and optimisation depth
KuberaMulti-entity, multi-currency wealth state
ShackademyCreator distribution around Voyant
Differentiation has to come from the model and the rate of learning.
## Slide 7
## The target sits in contested territory
```mermaid
quadrantChart
title Current product emphasis
x-axis Plan maintenance --> Continuous capture
y-axis Tracking and reporting --> Decision-model depth
quadrant-1 Integrated model
quadrant-2 Planning-led
quadrant-3 Point tools
quadrant-4 Tracking-led
ProjectionLab: [0.34, 0.88]
Voyant: [0.18, 0.94]
Monzo: [0.91, 0.27]
Kubera: [0.76, 0.53]
Asset Tracker target: [0.88, 0.90]
```
## Slide 8
## Monzo is the benchmark
# 15.2m
customers, with 10.4m using Monzo monthly.
More than 300k customers were investing through Monzo by FY2025.
### Its current boundary
External balances and transactions feed Trends. Investments, pensions and mortgages extend the product surface.
The provider account still anchors the model.
More account coverage still leaves the household model undefined.
## Slide 9
Shackademy packages financial decision support for £595
ReachUK finance educationCreator-led distribution
OfferVoyant + coursesCommunity and professional access
Signal£595 · 100 placesA priced, limited cohort
SetupManual licenceIssued within 48h; entering financial details takes time
Its advantage is distribution. Its friction shows what motivated users will tolerate.
## Slide 10
## ProjectionLab has moved; the thesis has to move too
### Founder test · spring 2026
The product foregrounded plan setup and simulation.
It did not replace my workflow for importing history, modelling financial reality and reading the resulting plots.
Monarch served the United States and Canada, which stopped the UK test.
### Product now
Actual-progress overlays, balance updates, UK tax presets and import plugins narrow that gap.
It remains the benchmark for planning depth.
Planning tools now track reality. The question is whether reality remains secondary to the plan.
## Slide 11
## Coding agents make one-user software worth testing
3 JulInteractive prototypeClient-side accounts, history and charts
7 JulMoney flowsMobile cards and Sankey view
11 AugHistory importsManual capital-flow records
30 AugFI progressDerived expenditure and trajectory
5–7 SepDecisionsHistorical projections and liquid runway
Five useful releases shipped in 66 days.
## Slide 12
## Four things can compound if they work in practice
CodeOpen sourceAGPL code can be inspected, corrected, exported and forked
PlatformShared monorepoComponents, deployment, tests and review are reused
DeliveryAgent-first loopContext, implementation, preview QA and review stay connected
ContextBroader digital twinFinance, commerce and household data connect only where useful
These are execution hypotheses. Defensibility depends on more than speed, and distribution depends on more than open source.
## Slide 13
## The project earns its next stage through evidence
UtilityOne live decisionThe model changes what I do
ReplaceSpreadsheet retiredHistory stays current in one system
ShareHousehold useMy partner understands it; another household onboards
VelocityFast and safeUseful changes land without growing rework
External users matter later. The first test is whether the product removes its own spreadsheet.
## Slide 14
## The thesis needs stop conditions
Decision
Keep building?
Review the evidence from use
Administration
The model costs more effort than the decisions are worth
Convergence
ProjectionLab, Kubera or Monzo serves the use case better
False velocity
Agent output creates rework faster than product value
## Slide 15
## Build the thing I need.
1. Use Asset Tracker for consequential personal decisions.
2. Replace the spreadsheet and reduce the cost of keeping history current.
3. Publish the code, model and evidence as the product changes.
4. Add household users, automation or commercial intent only when use justifies them.
Keep enough evidence to learn whether other people need it too.
---
Markdown index of this site: https://robbiepalmer.me/llms.txt