Enterprises Know Exactly What They Want From AI — And We’re Listening

The pace of change in the era of AI is accelerating at a velocity I never could have imagined. From the proliferation of the PC to the advent of the Internet, nothing compares in my 40-year business career.

What has not changed, however, is an important piece of wisdom for enterprise executives leading transformations: The best way for any company to ground its investments in reality and advance meaningful innovation is to source the path forward from customers.

While this assertion may appear obvious, I believe many technology vendors assume that customers “don’t know what they don’t know” about AI. This is patently false.

How do I know it’s not true? I have spent the last six months traversing North America, Europe and Asia listening to customers share what they have learned about AI, what they need from AI and what they expect their vendors to deliver.

In particular, I have talked to CEOs, CFOs, CIOs and other executives at some of the world’s largest and most consequential enterprises, from financial services and retail to technology and manufacturing. What has emerged from these conversations is a common set of AI requirements — a framework of what customers want and what technology vendors like my company, Anaplan, must deliver.

Here are those seven must-haves:

1. An Accuracy Imperative

Large language models, or LLMs, are predictive in nature, functioning through statistical pattern recognition. LLMs understand context, address uncertainty, interpret data and generate output. For all their power in generating answers, LLMs do not deliver the precision and mathematical accuracy required for standard enterprise use cases. For example, financial reporting requires 100% precision, fully traceable and auditable. LLMs are incredibly useful in delivering productivity, but if they are not coupled with deterministic AI for 100% reliable, traceable and reproducible calculations, they fail the enterprise accuracy imperative.