When International Business Machines Corporation rolled out the IBM 701 mainframe in the early 1950s, corporate America greeted it with great interest — and some skepticism.

The company’s first mass-produced mainframe was a leap forward when it came to processing data at scale, but it didn’t ship with an operating system. As a result, IBM had to develop customized software and unique training materials for each client. In 1952, the base monthly rental fee for an IBM 701 was around $8,100, around $96,000 today. (For context, gas cost 20 cents per gallon, and a new house was priced around $9,000.)

❝

I think the single best thing we can do as a profession is to help take the fear, uncertainty and doubt out of the problem.

— Jeremiah Stone, CTO, SnapLogic

IBM’s go-to-market strategy should sound familiar to anyone who follows emerging technology: to encourage adoption, then-CEO Thomas J. Watson Jr. built relationships with potential customers before offering them training seminars, workshops and hands-on demos.

Subscribe now

In turn, early adopters like Boeing, Prudential Insurance and General Motors became evangelists for the IBM 701, singing its praises at industry conferences and in trade publications, white papers and case studies.

Generative AI startups are at a similar crossroads today: their emerging technology could transform nearly any business process, but finding ways to integrate it into existing systems means entering uncharted territory.

Despite the hype cycle, a majority of enterprise customers “are spending 70-80% of their capacity maintaining legacy application infrastructure,” said Jeremiah Stone, CTO of SnapLogic. As a result, these companies don’t have a deep bench of talent to assess or integrate AI, and compliance and security are the two largest challenges facing AI-first founders.

To overcome these and other barriers, AI-first founders need to become educators and great listeners before they can ever act as salespeople.

In this episode, I interviewed Jeremiah to learn more about identifying and overcoming barriers to AI adoption, strategies for effective customer engagement and the importance of transparency and iterative experimentation.

“If you're not embarrassed by what you're showing, you're not doing it early enough,” he said.

“It is dangerous to be vulnerable, but in this space in particular with AI adoption, demystifying it, showing it not working great, explaining to people what's going on behind the curtain, I think the single best thing we can do as a profession is to help take the fear, uncertainty and doubt out of the problem.”

Episode summary

Security and compliance “are the two things that are challenging for adoption right now.”
Before selling AI solutions, founders need to listen to (and educate) potential customers.
“Enterprises are spending 70-80% of their capacity maintaining legacy application infrastructure.”
The risk of data breaches and bad customer outcomes makes enterprises cautious.

Focus on solving unique problems AI can address better than existing solutions.
Build trust by starting discussions with security, compliance, and risk frameworks.
Innovate openly and transparently to demystify AI for customers.
Ensure the team is prepared to answer typical security and compliance questions.

Maintain discipline in focusing on scalable solutions rather than niche problems.

“You have to create your own talent and upskill your data engineering team.”

“Experiment weekly on the things that look promising and treat it as an iterative process.”
Be transparent and involve customers early in your development process.