Operational Intelligence for a Payments-Orchestration Platform
Results at a glance
One live map of 31 people and 7 products, built from the tools they already used — naming the three people the whole operation quietly ran through, and a 'critical' product nobody was actually working on.
Named the three people the entire company quietly depends on
Found a "critical" product nobody was actually working on — invisible on their own board
Challenge
A 31-person company shipping 7 products, and the founder was still the only one who could see how it all fit together. The stuff that actually runs the business — which bank will approve which kind of merchant, why the payment setup was wired a certain way — lived in three people's heads. Four engineers were stretched across five "critical" projects, and one of those "critical" products had nobody actually working on it. Everyone there had quietly accepted running the company off a handful of memories.
Constraints
- check_circleHigh-stakes payments work — fraud limits and chargeback deadlines that shut your accounts down if you miss them
- check_circleAlmost nothing was written down — how the business works lived in people's heads, not systems
- check_circleSensitive payment data — any AI helper had to work without ever seeing the keys
Approach
I pulled their real company — people, products, the task board, the Slack threads — into one live map of who actually depends on whom. Then I picked the bottlenecks worth fixing first.
Architecture
- 01One live map of all 31 people and 7 products, showing who everything quietly runs through
- 02A product view that exposed the gaps — including a "critical" product nobody was actually working on
- 03A clear read on who's carrying too much, and who would get time back first
- 04A safe plan to add AI helpers without their payment secrets ever touching an outside service
Implementation
- 01Mapped all 31 people and 7 products straight from the tools they already use
- 02Named the three people the whole company secretly runs through — and exactly what stops if each one leaves
- 03Picked the six things to automate first — from the daily copy-paste between Slack and the task board, to the "go ask the one person who remembers" problem
Results
- check_circleOne live map of 31 people and 7 products, built from the tools they already used — naming the three people the whole operation quietly ran through, and a 'critical' product nobody was actually working on.
- check_circleNamed the three people the entire company quietly depends on
- check_circleFound a "critical" product nobody was actually working on — invisible on their own board
Business impact
- check_circleThe founder stops being the only person who can connect ops, payments, and the business
- check_circleSix concrete fixes ready to build next — a plan, not a slide deck
- check_circleA safe way to add AI without payment secrets ever leaving their control
Technologies
The research is public.
Benchmarks run on machines I own, measured against frontier systems — with the failed approaches reported next to the results. Read it before you take my word for anything.