←All work
01Multi-location photo studios · Bookkeeping automation
Turned a month's pile of POS, bank, and card statements into clean books in minutes.
Client buildDelivered · runs privately on the owner's own machine
Demo · synthetic data
The challenge
A multi-location, multi-company photo studio group lost days every month to bookkeeping prep: wrangling POS exports and a stack of bank and credit-card statements across several companies, approving transactions by hand with a bookkeeper, and still flying blind on how the business was actually doing month to month.
What shipped
- Statement upload with automatic parsing and duplicate protection
- Self-learning categorization engine that sharpens with every correction
- GST tracking plus a needs-review queue that surfaces only the uncertain items
- Four dashboard views with a universal day/week/month/quarter/year range and toggleable charts
- Insights tab with plain-English findings and an optional AI advisor on the owner's own key
- Accountant-ready exports, running entirely on the owner's machine — no cloud, no subscriptions
The stack
- Next.js
- TypeScript
- Claude (Anthropic)
- OpenAI
- Local-first (no cloud)
What changed
Monthly close went from a multi-day chore to minutes, and the owner can finally see the business clearly over time.
- Automation
- Local-first
- Dashboards
- AI
How we worked
01
Diagnose
Find the one thing costing the most time or money.
Constraint map02
Design
Map the system before building any of it.
Systems map03
Build & integrate
Ship the pieces and wire them together.
Working system04
Automate
Hand the repetitive work to the system.
Automations live05
Run
Keep it healthy and hand you the controls.
Monitoring + handover