THE OPS MANUAL
Architecting
Business Unusual
Deep dives into operational architecture, AI agent development, and data strategy. We share blueprints, methodologies, and real-world examples of what we build.
Building the Automated Consultancy: A 'Client Zero' Case Study
Most automated systems fail because they strive for 100% accuracy, often breaking down on edge cases. I decided to be my own Client Zero to prove there is a more pragmatic way.
I applied the same institutional precision I use for investment operations projects to my own back office. At Hamilton Parish, we built an AI orchestration layer that handles 95% of our expenses "hands-free," while maintaining a high-integrity audit loop for the remaining 5%.
Read the Full BreakdownHAMILTON_EXPENSE_AGENT.py
> INGESTING RECEIPT: Uber_Trip_NY.pdf
> EXTRACTING ENTITIES...
- Vendor: Uber [CONFIDENCE: 99.8%]
- Amount: $45.20 [CONFIDENCE: 99.9%]
> CHECKING POLICY COMPLIANCE...
- Flag: Match Project 'Alpha'
> ACTION: AUTO-APPROVE & SYNC QB
Fig 1. Real-time log from our internal Expense Orchestrator.