Aug 31, 2026 · 18 min
AI pilots fail when platforms ignore the business
EP 62: 95% of AI Pilots Fail. Here's Why | Mohamed Battisha, VP of Engineering at WEX
AI adoption depends less on adding models than on building the context, controls, and measurable business foundations that make automation useful.
- 1AI agents need semantic context, governed execution, and auditability beyond traditional reporting platforms.
- 2A crawl-walk-run approach helps organizations measure readiness, impact, and scalability before expanding automation.
- 3Business-aware engineers connect problem definition, data, tools, and infrastructure to outcomes that justify investment.
Don't miss
Battisha’s central advice is that AI and data engineers must become business-aware liaisons, connecting problem definition through infrastructure to measurable ROI.
The brief
Mohamed Battisha argues that reporting-oriented data platforms cannot support reliable AI agents without semantic context, governed execution, auditability, and clear business foundations.
Drawing on work at Wix and SDAIA, Battisha says AI initiatives should begin with a precise problem definition rather than technology adoption for its own sake.
Wix’s crawl-walk-run model moves teams from foundational capabilities to assistance and automation, using measurable impact and scalability to decide when to advance.
The episode’s sharpest career lesson is that AI and data engineers must understand the full chain from business problem to data, tools, infrastructure, and ROI.
Battisha’s broader conclusion is that rapid technical change only creates durable value when organizations can connect platforms and use cases to business results.
Featuring
Mentioned
Wix HarmonyListen to the full episode and explore every guest, topic, and moment on PodLume.