Agentic AI use-cases
Your edge in the world of AI is knowing technology but also in having the human big picture
When clients engage me on AI , they are increasingly looking at agentic AI use-cases. I try to make clear that they need to be very bounded in those use-cases. When I was building fully-automated trading systems, we spent a lot of time building circuit-breakers and firewalls as part of the delivery. Why? Simply because we didn’t trust the system not to be stupid, dangerous and market moving (nor did our regulators and compliance colleagues). So it goes with other use-cases that involve systems operating autonomously. As the famous 1979 IBM idiom goes:
“A computer can never be held accountable.
Therefore a computer must never make a management decision.”
By all means, use AI agents to automate tasks, or scan vast amounts of data for patterns and insights, trading strategies, cyber threats, KYC and onboarding automation, as these lend themselves to AI agents. BUT, so long as you are guarding your IP, business priorities and strategy, AI need not be involved.
Don’t forget, your edge is knowing technology but also in having the human big picture and nuance to make informed decisions about what your business needs to be doing next.