Enterprise AI has spent years in experimentation mode — demos, pilots, and prototypes. But according to Devavrat Shah, chief scientist of enterprise AI at Celonis, there’s a reason many projects stall: AI agents are trying to make decisions without understanding how the business actually works.
(Credits: MIT Technology Review)
Context-Aware AI and ‘Decision Intelligence’
Speaking on MIT Technology Review’s Business Lab podcast, Shah made the case for what he calls “decision intelligence.” AI agents can support individual tasks within larger decision-making processes, he argues, but they need to grasp the role each task plays and the business conditions surrounding it. “You can’t just do that in a vacuum,” he said.
Without the right operational context, agents risk overlooking important constraints, making costly mistakes, or burning excessive computing resources just trying to figure out what they need to know. A powerful model applying generalized principles to a specialized scenario will consistently come up short.
How Companies Should Start With Context-Aware AI
Shah’s prescription is practical: start with a specific business decision, identify its most difficult steps, and test whether AI can improve them. Measure the gains in speed and value, then decide whether the application is worth scaling.
The payoff could reshape how employees spend their time. Automating repetitive manual work and coordinating tasks across teams could free people to focus on complex strategic decisions — from finance and workforce management to marketing and sales. Realizing that benefit, though, will require investment in workforce education, so employees understand AI outputs and their implications instead of just handing decisions over to agents.
“Now, maybe [resource planning] should not take that much time, and people are actually making the right choices quickly,” Shah said.
Also Read: While Amazon Blocks AI Agents, Walmart Rolls Out Agent Gateway to Let AI Shop for You, Recursive Launches Flint, a Free Laptop Coding Model It Says Was Largely Built by AI
Source: MIT Technology Review — read the full story there.
