AI use cases.
Built for this industry.
These are not hypotheticals. Each use case documents a real AI workflow installed inside an equipment finance or banking operation — with the problem, the implementation, and the measurable result.
Foundation Protocol Implementation
0 → 44 triggers, 0 → 175 definitions, 0 → 42 agents in 8 weeks
"This is the foundational use case: it proves you can convert tribal knowledge into automation-ready specifications before you deploy automation. It's the missing layer most AI vendors skip."
Snapshot
- Client: Commercial loan servicer (equipment finance)
- Portfolio: $481.6M
- Engagement: 8 weeks, completed March 2026
- Starting point: Zero documented processes, zero automation
The Problem
- →Tribal knowledge dependency — no SOPs, trigger definitions, workflow diagrams, or terminology standards
- →Deal completeness failure — 10–12% of deals triggered rework loops; 35–55 minutes per rework cycle
- →Multi-system re-entry — agents toggled across 3 systems for basic inquiries; high duplicate entry for address changes and schedules
- →Visibility gap — no dashboards; leadership decisions driven by intuition vs. data
- →No customer portal — balance inquiries and payoff requests required live calls; payoff quotes tracked in spreadsheets
- →Platform debt — multiple LOS platforms and workflows across servicing clients; complexity compounded with growth
What Was Built
45 Task Specifications
Each task captured: trigger, inputs, owner, decision logic, outputs, system dependencies, and exception handling — built to the NOTBO standard across the full equipment finance lifecycle.
8 Production SOPs
CS & Payoff Inbox Maintenance, Assumption Request Processing, and all 5 DPD-stage Collections voice-agent scripts — each validated and ready for agent deployment.
44 Triggers Cataloged
EVENT, THRESHOLD, TIME, and EXTERNAL trigger types mapped across Origination, Underwriting, Funding/Docs, Accounting/Boarding, Titling, Collections, Servicing, and Insurance/Compliance.
Foundation Protocol Dictionary v2.0
175 definitions across 7 categories — acronyms, outcome codes, system/agent terms, collections/servicing terms, operations terms, compliance terms, and platform/vendor references. Sourced from transcript analysis, not invented.
42 Agents Designed
4 agents fully specified to production-ready requirements: Payoff Quote Processing Agent, Morgan (Collections Voice Agent), Customer Service Agent, and UW Data Aggregation Agent.
2 Production Dashboards
Operations Performance Dashboard and Portfolio Risk Monitor (PRM) — single-file HTML/CSS/JS, deployable immediately.
Results
Why It Worked
The protocol forced translation before automation design. Institutional knowledge became structured specifications. Undefined triggers became a cataloged inventory. Inconsistent vocabulary became a standardized dictionary. Every decision tied back to an SOP; every escalation tied to a named role and SLA. The dashboards weren't a separate initiative — they became possible because the underlying operational definition layer existed.
Client name suppressed per policy.