Are You Building AI With Your People — or Doing AI To Them?
AI adoption can fail long before the technology fails. Employees, customers and managers need to understand how AI is being used and where people remain accountable.
The technology is often the easy part
Choosing a model, licence or platform is only the beginning. Ask who will use the system, what problem it solves, what data it accesses, what behaviour changes and how people can challenge a bad result.
Shadow AI changes the starting point
Many organisations already have staff using ChatGPT, Microsoft Copilot, Gemini, Claude, meeting assistants, browser extensions and AI features inside existing software. First discover how AI is already being used before buying more.
Build with the people who do the work
Involve employees, managers, IT and security, data and privacy colleagues, and people affected by the output. Front-line employees often understand the workflow and its exceptions better than the people purchasing the product.
AI should know where its job ends
Define when a chatbot, recommendation, ranking, generated message or security signal must be escalated to a person. A human should remain accountable for consequential decisions and uncertain outputs.
Measure whether AI is actually working
Measure time saved, reduced errors, response time and useful information access alongside rework, staff frustration, security incidents, customer complaints and increased software cost.