The Uncomfortable Reality of AI for IPO Preparation

AI for IPO preparation can cut the workload by finding software overlaps and handling local regulations, according to Bret Taylor at Sierra and one panelist at HumanX 2026. StackIQ found 35 overlaps across one company's software stack, the panelist noted.

For finance leaders staring down an IPO, pointing generic software at a broken reporting process usually adds an expensive bottleneck. Taylor recalled the opposite experience when his team went live with a Spanish bank and had to work through EU and local AI regulations.

What makes IPO preparation hard for finance leaders?

One panelist noted that organizations often pay $600 per employee per month for AI licenses with no idea what the return is. Another noted that 94% of organizations see no meaningful bottom-line impact from their AI investments. Companies hire an expensive chief AI officer with no internal data to work with, then pay consultants millions for use case advice with no execution layer.

Overlap is where the money hides. One panelist recalled that StackIQ found Workday and Smart HR running side by side at one company, two HR systems with 68% feature overlap. A customer signed in March was already able to consolidate support tools that had been billing it for overages.

Where does AI for IPO preparation help with data collection and analysis?

At Verkada, a late-stage camera company, the technical leadership team found a number of things only they could do because of all the video they were sitting on, another panelist noted. CFOs at several Fortune 100 companies cared about the solution that came out of it.

Taylor said one medical devices company he works with has 40 call centers and is consolidating around one agent.

How do you get through compliance and security?

Getting agents into production means surviving the regulatory maze. When Taylor's team took one of the largest Spanish banks live, they had to navigate a web of EU and local AI regulations.

Larry Li at Amino Capital called security the biggest hurdle, especially because humans cannot follow agent-to-agent communication. There are a lot of security issues out there, he said, and agent-to-agent talk is where people lose the thread.

How fast can teams go live?

Taylor said customers such as Nordstrom went live in four weeks, and Cigna, a Fortune 20 healthcare payer, took two months. He predicted that transforming part of a business in days, weeks, or hours, instead of through 12-month implementations, will change software.

When an agent answers a provider-facing or patient-facing phone line, Taylor estimated that technology is maybe 40% to 50% of the problem.

What are the limits of AI in IPO preparation?

One panelist observed that some teams spin up about seven agents, go to bed, and come back to find they have forgotten what two of them were doing. They kill those and start two more. The panelist asked what the actual job to be done is, and whose need the team is trying to serve.

Taylor's work with the Spanish bank shows that fast deployment and strict compliance can coexist. If your AI strategy is buying licenses without checking for overlap, you are likely part of the majority seeing zero bottom-line impact, and AI for IPO preparation will not fix that on its own.

To hear more from the people putting AI into production, see the HumanX 2027 program preview.