Production-Grade AI: From AI Activity to EBIT

  • Simon

This article is based on Chapter 1: Reality Check from Production Grade AI If you’re feeling bruised by endless AI pilots, endless “activity” and zero change to the balance sheet, you’re not alone. But there’s a path to break the cycle.

Production-Grade AI: Building the Agent Hub

  • Simon

This article is based on Chapter 5: Building the Agentic Ecosystem from Production Grade AI The first wave of AI wins looks great—until it doesn’t. Five assistants all call finance differently. A minor vendor outage takes half your workflows down. No one can explain why an agent did what it did, or how much that decision cost. That’s not a model problem. It’s a systems problem. The answer is an agent hub.

Production-Grade AI: Boardroom to Bottom Line

  • Simon

If your organisation has a dozen AI demos and not a single line on the P&L to show for it, you’re not alone. The past two years have produced a glut of proofs‑of‑concept, assistants in side tabs, and model experiments. What separates leaders now is not model access but operating discipline: turning experiments into production overlays on systems of record, with stage gates, deprecation, and KPIs that speak the language of the balance sheet.

Production-Grade AI: From Copilots to Controlled Autonomy

  • Simon

Most teams start with copilots in a side tab. Most value shows up when AI sits in the path of execution. The shift that matters for operations and product leaders is from “helpful on the side” to “overlays inside the system of record (SoR), with clear deprecation gates.” That’s how you change behaviour, measure impact, and scale safely.

Production-Grade AI: Governing Autonomy Without Killing Speed

  • Simon

Most AI programmes stall not on modelling, but on oversight. Risk, Legal, and Compliance want control; Product and Engineering want speed. You can have both—if governance is operationalised into the delivery path instead of bolted on at the end. The shift is from document‑led sign‑off to evidence‑by‑design: risk tiers that drive default controls, clear decision rights, progressive rollout by blast radius, and audit artefacts captured automatically.

Implementing Software

  • Simon

Software projects fail when they solve the wrong problem, or no one uses them. If a project delivers there should be so much value that cost overruns aren’t the issue.

How to select software products

  • Simon

Selecting a software product is like choosing a house for your business to live in, except it’s more expensive and the risks are higher.

Revolutionise Your Content Management: Building Custom Apps for Contentful

  • Ryan

Integrating with Contentful is a quick and easy way to give non-developers Control over site content. But what if you want more?

Integration Testing! But what about emails?

  • Matt

Order success emails, Passwordless logins, testing rendering logic, how on earth can I test this?