AI, Technology, Transformation, and Inspiration

Writing about AI, technology and commercial growth: what worked, what failed, and what helped me move from pilot to production.

Nathan Petralia at HKU

Nathan Petralia

I have spent two decades leading digital and commercial programs across APAC. Today I build AI products and leverage AI across consulting, practice building, go-to-market, commercials, delivery governance, and operational leadership.

About me →

Recent Posts

The Knowledge Work Agent Engine: A File-Based Stack for PM, Leadership, and Marketing (Not Just Code)
Career

The Knowledge Work Agent Engine: A File-Based Stack for PM, Leadership, and Marketing (Not Just Code)

The same session-continuity engine that ships software can run initiatives, decisions, and content. Maps memory, voice, and routing to Agile, Jira, Confluence, RACI, and RAG—with a replication kit an AI can execute.

Continue reading →
Leadership and Decisions With an AI Session Engine (Purpose, Dissent, and Audit Trails)
Career

Leadership and Decisions With an AI Session Engine (Purpose, Dissent, and Audit Trails)

Simon Sinek's Why-How-What, Drucker's decision discipline, and RACI meet applied AI. Leaders keep accountability; the file-based engine holds purpose, dissent, and decision records agents need at session start.

Continue reading →
Marketing and Voice at Scale With a File-Based Agent Engine (Systems, Not Style PDFs)
Commerce & Marketing

Marketing and Voice at Scale With a File-Based Agent Engine (Systems, Not Style PDFs)

Brand voice fails when it lives in a PDF nobody opens. This playbook maps Sinek's Why-How-What, voice-as-system governance, and content-batch routing to produce consistent, high-volume marketing with minimum rework.

Continue reading →
Project Management With a File-Based Agent Engine (Not Another PM Tool)
Career

Project Management With a File-Based Agent Engine (Not Another PM Tool)

Agile, Scrum, Jira, and Confluence already own execution and narrative. This playbook shows where a file-based agent engine fits—iron triangle tradeoffs, RAG, RACI, RAID, and applied AI without pretending chat is a program office.

Continue reading →
Measure Your Cursor Harness — CSV, CI, and OpenRouter Dollars
AI & Building

Measure Your Cursor Harness — CSV, CI, and OpenRouter Dollars

Do not build Phase 2 orchestration until Phase 0 data says so. Layer 4 feedback — CSV, footer Agents line, eval gate — plus weekly OpenRouter checks beat benchmark leaderboard anxiety.

Continue reading →
Agent Harness Memory Loop — Four Tiers, Feedback Loop, and Load Gates
AI & Building

Agent Harness Memory Loop — Four Tiers, Feedback Loop, and Load Gates

External memory is four tiers in practice — short-term, operational, evergreen, and a feedback loop hardened into rules and footers. The harness gates when each tier loads so you keep control without token bloat.

Continue reading →
You Already Have an AI Harness in Cursor (Without LangChain)
AI & Building

You Already Have an AI Harness in Cursor (Without LangChain)

Terminal-Bench harnesses look like separate products. On a production Shopify app I already had subagents, CI gates, and session rules. You keep model and mode control — the harness supports routing, tests, and memory gates, not autopilot.

Continue reading →
CursorBench 3.1: Fable 5 Tops the Chart, but Composer 2.5 Wins the Budget
AI & Building

CursorBench 3.1: Fable 5 Tops the Chart, but Composer 2.5 Wins the Budget

Anthropic's Fable 5 leads CursorBench 3.1 at 72.9%, but at $18 per task and 76 steps. I read the table for score per dollar, tokens, and steps, and where open models land.

Continue reading →
Beyond Headroom: What I Tried to Save Cursor Tokens, What Failed, and What I Use Now
AI & Building

Beyond Headroom: What I Tried to Save Cursor Tokens, What Failed, and What I Use Now

I ran Headroom, built a 300-line proxy, wired a Cloudflare tunnel, and added RTK. On my Cursor + OpenRouter workload the dollars did not move. Here is what is worth doing instead.

Continue reading →
Browse all posts →