Case Study
How our automations actually work — from the inside.
Real systems we've built, documented end to end. Architecture, controls, results — including the ones that didn't pass.
Case study
01
Market Intelligence Crew
8-agent CrewAI system for market intelligence — and why its most important output was “no”.
- crewai
- lightgbm
- multi-agent
02
AutoClaim
Operations layer for vehicle insurance claims — 5 stages, 7 specialist agents, and one rule: every unclear case goes back to a human.
- operations-ai
- hitl
- insurance
03
Fershteyn Law
AI visibility for a Brooklyn estate planning firm — citation rate across Gemini, Perplexity, and ChatGPT went from 16.1% to 62% in 8 weeks (3.9× lift).
- aio-geo
- schema-markup
- legal
04
AI-Avatar Content Engine
A photorealistic digital-twin and cloned-voice pipeline that turns one recording session into ~90 published videos a month across nine platforms — daily, compliant, for a firm in a strictly regulated market.
- ai-avatar
- video-pipeline
- legal
05
Profitec Operations CRM
A configurable operations CRM that routes records through validation, AI classification, confidence thresholds, human review, and an audit trail — shown in an interactive demo.
- operations-crm
- controlled-automation
- hitl