Deterministic Porting & Agentic Engineering: Artifact Repository
Overview
This repository contains a collection of presentations, infographics, and mind maps explaining the “Provable Porting Engine” developed by Blair Nilsson to migrate legacy SAS databases and Excel spreadsheets at scale. The materials contrast Agentic Engineering with vibe coding, illustrating how deterministic orchestration, strict LLM bounding, and mathematical testing can turn a multi-year migration project into a provably accurate, overnight automated run.
🎥 Talks & Videos
Click a thumbnail to watch on YouTube.
| Video | Talk |
|---|---|
| AI Agents Transform Legacy Code: How to unscramble an egg Agentics NZ — turning software back into provably correct specs. |
|
| Blair Nilsson’s Provable Porting Engine Agentics NZ — migrating legacy SAS & spreadsheets at scale. |
📊 Presentations (Slide Decks)
Click a thumbnail to open the full deck.
| Slides | Presentation |
|---|---|
| Agentic Porting at Scale “Unscrambling the Egg.” The Stats NZ case study — provable migration of legacy SAS and Excel systems via deterministic AI orchestration, transforming 20 years of technical debt into modern infrastructure. |
|
| Provable AI Porting The Business Value of Provable AI Porting. The executive / business case for reclaiming technical sovereignty from legacy monoliths. |
|
| Unscrambling the Monolith The comprehensive explainer — migrating legacy SAS & Excel to modern stacks via agentic engineering. |
|
| The Provable Porting Engine The technical deep dive — migrating legacy SAS & spreadsheets at scale via agentic engineering. |
🖼️ Infographics
Click a thumbnail to open the full-resolution image.
| Visual | Description |
|---|---|
| The Anti-Hallucination Loop A provable path for legacy code porting: the forward pass (deconstruction), the backward pass (verification), reassembly, and mathematical testing — trapping AI hallucinations to a <1% failure rate. |
|
| Provable High-Accuracy Code Migration Process Why dumping 200k zipped XML files into an LLM fails, and how the deterministic slicer cuts code into sub-100-line fragments for 100% accuracy. |
|
| From Messy Formulas to Clean Python The dependency-graph transformation: messy Excel R1C1 formulas are deterministically parsed into clean Mermaid node charts that serve as the prompts for the AI. |
|
| The Great Migration: Traditional vs. Agentic Porting A visual comparison of manual human porting versus AI agent porting across cost, speed, and accuracy. |
|
| The New Critical Path: How AI Flipped the Migration Timeline How the core bottleneck has shifted away from writing code and toward downstream systems, environment setup, and organizational approvals. |