Worldwide Planning Intelligence
I built WPI for multi-program planning at scale: scored risk register, dependency tracking, executive reporting, and local LLM triage. Published as a white paper, not a hosted product.
Risk Register
Scored, tiered P0–P3 register. Filter by tier, category, or owner.
Guided Intake
AI asks structured questions and produces a scored risk record from plain-language answers.
Dependency Log
Capital, resource, and cross-program linkages tracked in one view with navigation between entries.
Executive Report
Weekly report in pyramid format: decisions at the top, analysis below. Built for director review.
LLM Benchmark
Five local models evaluated on triage quality, speed, and rubric score before production selection.
Benchmark results
I ran five local models against 82 risks with a structured rubric: tier accuracy, field completeness, mitigation quality. Llama 3.1 scored 76% at 7.5 seconds per call. Gemma 3 hit 73% at 46.7s. Mistral 66% at 26.2s. Qwen 2.5 56% at 38.6s. The white paper walks through how to reproduce this in your own environment.
Local vs. cloud
Local inference was $0 per call on hardware I already owned. Cloud API cost was about $0.009 per call. Cost matters, but privacy mattered more. Every cloud call sends program data off-premises. For companies with OEM commitments or competitive IP, that is a governance call.
What held up in testing
- Smaller 8B models beat larger 32B models on speed-adjusted triage quality
- RAG with historical decisions improved grounding in manual review
- Guided intake cut risk creation from about 15 minutes to about 3
- One register with tier filtering replaced three separate tracking documents
This is not a hosted product. If you want to walk through the setup, reach out. For tools you can run in the browser today, see the Program Risk Register, Roadmap Builder, and the full tools catalog.
Built by Arsenii Samoilov, a Senior Technical Program Manager with 19+ years at Intuit, Atlassian, Adobe, Salesforce, Roku, and Apple.
Read the white paper →