The AI in Business Podcast Operationalizing Customer Service at Scale with Outcome-Driven Agentic AI - with Craig Walker of Dialpad
15 snips
Mar 9, 2026 Craig Walker, Co-founder and CEO of Dialpad, a serial entrepreneur applying AI to customer service and voice workflows. He recalls early chatbot failures and why prior tech fell short. He explains shifting to AI as an agent that handles repeatable tasks, strategies for pilots and scaling, cleaning knowledge bases, ticket analysis, and using AI for real-time coaching and analytics.
AI Snips
Chapters
Transcript
Episode notes
Why Prior AI Failed CX Trust
- Earlier generations of customer service AI created distrust because they aimed to keep customers from humans and produced awful experiences.
- Craig Walker says modern AI shows green shoots of being genuinely useful and scalable where prior models failed, changing market expectations.
Run Small High-Confidence Pilots First
- Start with a unified vision and small proofs of value by running a demo or POC with a handful of agents rather than ripping out your whole org.
- Craig Walker recommends five agents in a pilot so you can prove combined AI+agent productivity before broad rollout.
Fix Your Knowledge Base Before Training AI
- Clean and maintain your knowledge base because agentic AI will rely on those help articles as gospel.
- Craig Walker warns feeding stale help-center articles into the AI will train it on wrong answers and cause real problems.