Preparing for the AI Agent Revolution with John K. Thompson

[@markstouse4635]

How do you square the circle between agentic AI and the fact that most companies have very significant data integrity issues? Thanks

[@johnthompson6237]

Hi Mark, Great question. Thanks for jumping in.
GenAI and all AI models require good, clean, accurate, and relevant data for training, fine tuning, and on-going calibration of models. That does not change.
Agentic AI where agents can have probabilistic and deterministic models as part of each agent and agentic workflows are, in the main, working on transaction level data and decision level information.
When agents and agentic AI is executing processes, the data that is in process needs to be accurate and relevant, but most of that data is going to be resident in source systems or systems of record, which, by its nature should have a level of quality that is within defined tolerances.
So, in the end, training and fine tuning models does not change with agentic AI. Still the same things that we have done for decades.
Agentic AI is operating at the process level and the data in those processes needs to be as good as it is today. No real change there either.
Did I answer the question, or miss the mark? No pun intended.