Keith: What key governance steps should businesses, large and small, take before moving into production? Ed: Like any software, AI must go through security reviews, SDLC processes, observability, and metrics. All boxes must be checked. But AI adds further risks.
Bias, inference abuse, unauthorized access by chatbots, etc. need to be monitored. Let's think about the worst case scenario. What is the explosion range? Can you pull the plug quickly if necessary? Keith: And sometimes there's pressure from the business side to just ship it. Edit: That's right.
Influencers make it look easy, but MVPs only do 80% of the work. The remaining 20% (security, observability, scaling) is the real work. Cannot be skipped. Keith: So do you recommend that small businesses pursue small profits first? Ed: Exactly. I have strong opinions here.
Most employees don't use AI for anything other than creating summaries and help. But with tools like NotebookLM and DeepResearch, you can do weeks' worth of research in minutes, greatly increasing your productivity. Every employee needs to find their own AI win. It builds confidence and literacy.
Keith: So the goal is for humans to get used to the AI and find wins organically? Edit: That's right. This is the “Hi-Hi-Walk-Run” model. You have to play around with tools, push boundaries, and learn your limits. This allows you to build more complex workflows.
