Practical writing on using AI well

I'm Nathan Petralia, operator and writer on enterprise AI, file memory, and getting agents to production.

What worked, what failed, and what helped me move from pilot to production.

New posts land on the blog. One weekly email rounds them up if you subscribe.

Nathan Petralia at HKU

Nathan Petralia

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

About me →

Start here

New here? Pick one path. Each links to a guide or series hub.

Browse all writing →

Reading paths

Multi-part guides. Start at the hub, then read in order.

Latest

New this weekA hand annotating a printed line item on scope pages in a Sheung Wan agency meeting room, in warm late afternoon light.
Strategic

How GenAI Is Resetting Agency Pricing

Best forAgency and client-side marketers who need to move AI-era scopes from hours to deliverables they can defend, with QA gates and verification lines priced in.

GenAI broke the hours-based retainer from both sides. The reset that works prices deliverables and outcomes with QA gates, revision SLAs, and verification lines.

Continue reading →
New this weekFutian warehouse packing desk with two phones and carton stacks suggesting multi-channel APAC commerce.
Hybrid

Shopify Agentic Commerce for APAC: WeChat, LINE, and Grab Without Duplicating Catalogs

Best forAPAC merchants and program owners running Shopify plus super-app channels who need agent-ready catalog sync without triple data entry

Shopify's agentic commerce pitch is US-shaped. APAC operators need one catalog feeding WeChat Mini Programs, LINE carts, and Grab APIs, not three manual spreadsheets.

Continue reading →
Two managers in an office pantry reviewing a printed chart, illustrating the gap between AI deployment and measurable goals.
Strategic

57% Deployed, 11% Hit Their Top AI Goals: What Kyndryl's 2026 Report Measures

Best forProgram leads and CX owners who need primary-source stats to interpret deployment vs outcomes and reframe AI ROI around people readiness, not license counts

Kyndryl's 2026 People Readiness Report finds AI deployed in 57% of enterprises but only 11% hitting their top two goals. The gap is people and program design, not model access.

Continue reading →
Cinematic 16:9 of a retail receipt and loyalty card on a brushed steel counter under warm spotlight.
Strategic

CX Metrics Agents Cannot Fake (and the Ones They Will)

Best forCX and program leaders choosing scorecard metrics when AI assists service, content, and reporting

Agents optimize what you measure. Vanity CX scores and synthetic survey text are easy to game. Operational metrics tied to customer behavior are harder.

Continue reading →
Hong Kong night market shoppers at adjacent stalls, one paying by phone, illustrating two social commerce habits in one real place.
Strategic

WeChat Mini Programs vs Instagram Shop: Where Social Commerce Still Needs a Human Program Owner

Best forCommerce and CX leaders choosing between APAC super-app mini programs and Western social shop surfaces who need a governance model, not another platform pitch

WeChat Mini Programs and Instagram Shop both promise in-app checkout, but they govern data, pace, and accountability differently. Neither replaces a human program owner who owns the calendar, the consent layer, and the handoff to CRM.

Continue reading →
Cinematic 16:9 wide shot of a construction site at dusk with an illuminated site office trailer.
Strategic

Digital Transformation Is a Repo With Agents and a Bridge File

Best forProgram executives tired of transformation theater who want delivery mechanics that connect strategy decks to weekly agent-assisted execution

Transformation programs fail when intent lives in decks and delivery lives in chat. A repo, governed agents, and a Bridge file align purpose with weekly execution.

Continue reading →
Browse all writing →