Everyone Wants to Build AI. Who's Going to Maintain It?

Why It Matters

The choices organizations make now—whether to build or buy AI—will determine their ability to scale safely and avoid costly long-term technical debt, affecting productivity, compliance and competitive advantage across industries.

Summary

Speakers identified five core tensions shaping enterprise AI adoption—pilots versus production, speed versus trust, restructuring versus workforce readiness, build versus buy, and control versus visibility—and argued these trade-offs are driving strategic uncertainty. The conversation focused on the build-versus-buy dilemma, warning that while organizations can now build in-house AI, maintaining, scaling and governing those systems may create persistent operational headaches. Panelists cautioned that bespoke, hastily developed AI tools risk creating a new wave of “sloppy” tech debt—analogous to prior legacy problems—because companies may lack the resources or discipline to sustain custom models long-term. They also noted the ongoing value of point solutions, which encapsulate maintenance and reduce organizational burden despite the allure of internal ownership.

Original Description

Build versus buy sounds like a business school debate until you realize the travel industry is still digging out from years of tech debt it barely managed to offload to the cloud. In this clip from the Skift Travel Podcast, Seth Borko and guest host Sean O'Neill break down the five biggest AI tensions facing the industry right now and why the question of who builds what in-house could define the next decade.

Because vibe-coded AI tools shipping from the front desk might solve a problem today and create five more by 2035.