Why companies are treating AI as a strategic partner rather than a passive technology, and how to avoid an ‘AI hangover’

Building operational AI at scale requires more than just raw algorithmic power, business and tech experts said at Fortune Brainstorm Tech.

In corporate America, AI has moved far past the hype cycle into practical implementation. Autonomous agents are handling complex, real-world tasks on behalf of companies that go beyond simple data insights.

But building operational AI at scale requires more than just raw algorithmic power. At Fortune Brainstorm Tech this week, executives from third-party logistics giant C.H. Robinson, Gap, AI lending platform Upstart, and object storage firm MinIO argued that true enterprise success requires treating AI as a strategic partner. This involves implementing something like a digital supervisor that acts like a centralized safety layer above AI models to enforce guardrails, ensure compliance, and guarantee error-free results for enterprises.

For C.H. Robinson, this shift has rewritten their rules of logistics. The company is using AI to manage the thousands of natural-language emails customers send daily, which previously could not be automated effectively with traditional software code, according to Chief Technology Officer Mike Neill. By deploying a tool known as an AI classifier to instantly identify customer intent, C.H. Robinson has slashed response times to as little as 32 seconds, he said. This automated pipeline handles everything from booking orders to securing appointments, driving what he called massive efficiency gains.

But for AI agents to become truly production-ready, they must evolve to becoming systems with long-term memory and contextual awareness, executives said.

“AI agents must remember customer preferences,” said MinIO co-founder and co-CEO Garima Kapoor. But keeping that memory in expensive GPU hardware isn’t financially sustainable for growing companies, she said. She noted rising demand for new software that moves AI memory off expensive chips and into cheaper storage. This lets agents smoothly remember past conversations instead of resetting every time.

“Generative AI is more about prompting,” Kapoor said. ‘Where agentic workload is taking decisions based on the data that you are seeing flowing through.”

Technical infrastructure, however, is only half the battle The human element remains a steep hurdle. Sven Gerjets, chief technology officer at Gap, warned that technical training alone can backfire. Employees who received only basic prompt engineering training often experienced what he called “AI hangover,” which he described as a sharp drop in workforce confidence when the technology failed to instantly solve complex, multi-layered retail problems.

To combat this, Gap is pushing for a cultural mindset shift. Company growth succeeds only when employees stop viewing AI as a passive software tool like Excel, and begin treating it as an active digital coworker, Gerjets said.

If “you look at this as a partner,” he said, “you actually can get a lot more out of it because you’ll ask it [questions like], ‘I don’t know what to do?’”

More from the 25th annual Fortune Brainstorm Tech conference:

Anthropic’s Boris Cherny, creator of Claude Code, says there are days he manages tens of thousands of AI agents at once

The AI industry spent years chasing bigger models. Now it’s chasing efficiency

‘Not an Allbirds Moment’: Xbox’s new CEO says she is grounding the console in gaming roots not AI

This story was originally featured on Fortune.com