
Cognizant is at Ai4 for a short list of working conversations about moving enterprise AI from pilot into production. Sushant Warikoo is on the main stage, and Imran Masood, who runs operations, is on the floor all three days.
MIT put a number on it. The models are extraordinary, so capability is not the constraint. The operation around the model is. A pilot demos beautifully, hits its first real exception in production, and quietly gets rolled back, because the roles, the supervision, and the retraining were never rebuilt to carry it.
Source: MIT study on enterprise AI adoption.
A procurement operation we run for a global manufacturer. Live in production, and paid on the outcome, cost coming down against a baseline, not on headcount.

Who owns the exceptions, who checks the agent, how the work gets retrained when policy changes. The layer most programs skip.
Each workflow maps to a level of autonomy and moves up only when the evidence earns it, so cost-to-serve comes down instead of doubling.
One of the world's largest professional services firms, running operations for global enterprises in production.
The AI initiative that looked better in the demo than it does in production. Real numbers help, but the shape of the problem is enough.
Imran walks it across the autonomy levels we use to run this work: assisted, supervised, autonomous with exceptions, autonomous within bounds.
Where it is stalling, and what evidence would move it up a level. Yours whether or not we ever work together.

Runs operations for Cognizant. His day job is the part of AI nobody demos: the exceptions, the supervision, the retraining, and what it takes for the numbers to hold once the pilot is over. On the floor all three days.
Named one of RETHINK Retail's Top AI Leaders in Retail. Speaking on the Ai4 main stage on industrializing AI in retail.
Ai4 main stage · Aug 4-6
August 4-6, Las Vegas. Pawan Dasaik coordinates the calendar. Reply to his note with the time that works and he'll lock your slot.