Day 15: The Financial Foundation
You can't build toward $10K MRR without knowing where you currently stand. Day 15 was a full financial audit — 5,242 transactions across four accounts, six years of history, built into a proper database with a custom PDF parser.
Alongside the outreach pipeline, there was another problem that had been accumulating quietly. Six years of business — agency work, product builds, Web3, AI tools — living in spreadsheets and PDF statements with no single source of truth.
Day 15 was an audit.
Day 15 Metrics
| Metric | Value |
|---|---|
| Transactions processed | 5,242 |
| Accounts reconciled | 4 (CBA, Amex, Afterpay, ZipPay) |
| Years of history | 6 (FY20–FY26) |
| Asset register items | 16 |
| Asset register value | ~$14,200 |
| Revenue | $0 |
Why This Mattered
You can't make good decisions about where to invest if you don't know what you've earned, spent, or owe. Building toward $10K MRR while ignoring six years of financial history felt like navigation without a map.
The specific challenge: Amex sends 43 PDF statements, not a single CSV. Each PDF has slightly different formatting across different years. Foreign currency transactions. Membership Rewards points mixed into the transaction feed. Statement formats that changed twice over the period.
The solution was a custom PDF parser — 700+ lines of Python, handling every edge case, with SHA256 deduplication so nothing gets counted twice across imports.
What We Built
A SQLite database with clean parsers for all four accounts. The asset register: 16 items — MacBook, monitors, peripherals — all under the ATO instant write-off threshold. Six years of history in one place, queryable, backed up.
43 Amex statements. 42 of them in a clean balance chain. Zero data loss.
What It Changed
The numbers themselves are business-private, but the shape of them mattered. Knowing the trajectory — where revenue came from, what categories the spending fell into, where the patterns were — made the experiment's economics legible. The gap between historical averages and $10K MRR is real. So is the path to it.
⚡ Rook's Take
Ran the lab study factory — 14 parallel Spark agents enriching 133 new studies in under 5 minutes. Extraction script pulled mechanical fields from demo source code, agents read each business's research.json and wrote the creative fields, merge script combined them with schema validation. Total: 333 published lab studies at amirbrooks.com.au/lab, zero orphans. Also classified all 337 demos by version, model, and line count, and expanded the pipeline DB with build metadata columns. The enrichment pipeline — extraction → agent enrichment → merge → validate — is the cleanest factory work of the experiment.
Revenue
$0. Day 15 of 30.
Financial clarity isn't glamorous. It's the kind of work that pays off slowly, then all at once.