AI and the Future of Work
380: Customer Service's AI Shift: Zendesk CTO Adrian McDermott on Deterministic AI and Context Engineering
Why It Matters
The conversation reveals how AI can transform frontline support functions without necessarily eliminating jobs, offering a roadmap for organizations navigating automation and workforce changes. As AI adoption accelerates across enterprises, understanding deterministic AI and context engineering helps leaders implement trustworthy, scalable solutions that improve both employee experience and customer satisfaction.
Key Takeaways
- •Zendesk shifted from product-led growth to AI-driven strategy.
- •85% of developers use AI coding assistants weekly.
- •Deterministic AI reduces hallucinations, ensures reliable customer responses.
- •Co‑pilot AI handles intent, sentiment, and task automation.
- •AI adoption grew Zendesk’s revenue to $200 million in 18 months.
Pulse Analysis
In this episode, Zendesk CTO Adrian McDermott recounts the company’s journey from a lean, product‑led startup to a market leader embracing AI‑driven customer service. He highlights how the firm’s early focus on "convention over configuration" created a scalable platform that now powers billions of support interactions. The discussion underscores why AI matters: it transforms legacy contact‑center workflows, reduces service debt, and aligns with broader trends in generative AI, making Zendesk a case study for enterprises seeking to modernize support operations.
McDermott dives deep into the technical side, explaining Zendesk’s push for deterministic AI to tame the unpredictability of large language models. By integrating retrieval‑augmented generation (RAG) with their massive knowledge‑base ecosystem, the company delivers accurate, context‑aware answers while minimizing hallucinations. Their Co‑pilot suite goes beyond simple text completion, detecting intent, sentiment, and even executing back‑end tasks on behalf of agents. This context engineering not only boosts agent productivity but also standardizes training, delivering consistent, high‑quality service across B2C, B2B, and employee‑help desks.
The business impact is evident: AI‑powered features have driven Zendesk’s revenue to $200 million within 18 months, and 85% of its developers now rely on AI coding assistants daily. While automation reshapes entry‑level support roles, McDermott stresses a shift rather than wholesale job loss, with new opportunities in AI supervision, data labeling, and advanced workflow design. For leaders, the takeaway is clear—invest in deterministic AI, prioritize knowledge‑base integration, and use AI to augment, not replace, human agents to sustain growth and customer trust.
Episode Description
Send a text
Adrian McDermott is Chief Technology Officer at Zendesk, where he leads the company’s product management and engineering teams and helps shape the technology behind one of the world’s most widely used customer service platforms. He joined Zendesk in 2010 and has played a key role in guiding the company’s product and platform strategy as customer experience continues to evolve in the age of AI. Drawing on years of experience building enterprise software used by service teams around the world, Adrian brings a thoughtful perspective on how AI can help organizations deliver better customer service while allowing people to focus on the work humans do best.
In this conversation, we discuss:
How customer service evolved from a cost center with rigid scripts and binders into a strategic function where technology helps teams deliver better experiences.
Why customer service leaders shouldn't fear automation — and why everyone has a "service debt" that AI can finally help pay down.
The shift from traditional contact centers to AI-enabled service platforms that help companies respond faster while improving both employee and customer experience.
Lessons Adrian learned scaling Zendesk from a small product team to a global platform serving 100,000 customers and how product-led growth shaped that journey.
The critical challenge of moving from non-deterministic, creative AI models to deterministic, reliable solutions necessary for enterprise trust and safety
The future of context engineering and why the next major leap in AI won't be about superintelligence, but about building systems that capture and act on the knowledge created in every customer interaction.
Resources:
Subscribe to the AI & The Future of Work Newsletter
Connect with Adrian on LinkedIn
AI fun fact article
On How the impact of the pandemic on leaders, culture, and the evolving nature of work
Comments
Want to join the conversation?
Loading comments...