Insights
Perspectives on agentic AI, conversational AI and enterprise adoption — from the team that builds it.
Conversational AI Is an Omnichannel Problem
Wiring tools directly to a conversational agent is tempting because it makes the decision layer disappear — but pricing and eligibility decisions still need to trace back to a rule, a version, a timestamp. The discipline this calls for is the one omnichannel programmes already built.
Read articlePrompt Debt: The Software Engineering Problem Hidden in Your AI Stack
A system prompt plays the same central role in a mission-critical conversational AI agent as the domain model in complex software — and like every domain model, it drifts silently. Governing the prompt layer takes real engineering discipline.
Read articleThe Four Tiers of Agentic AI
Knowledge Chatbots, Agentic Co-workers, Autonomous Agents and Mission-Critical Agentic Software are four distinct tiers — and knowing which one fits your situation is worth more than any particular technology choice.
Read articleMy Agentic AI Close Call
An agent was about to send confidential architecture data to a third-party server — without breaking a single rule. Why agentic AI moves governance from the policy level to the architecture level.
Read articleThe Reality of Shadow AI
Shadow AI isn't a discipline problem — it's a structural gap between where productivity is happening and where governance exists. What that gap looks like in practice, and the first step toward closing it.
Read articleThe AI productivity gap is a structural problem
Your colleagues are already using AI — but the promised productivity gains haven't materialised at the organisational level. The gap isn't access to tools; it's the absence of structure.
Read article