Andrew Lavers
Andrew is a principal software developer at Zendesk working out of the Montreal office. Since 2015 he's been working on the business-to-user conversation platform that began as Smooch.io (acquired by Zendesk in 2019). He has 20 years of experience in software development, previously working on identity and access projects at Microsoft, and tech support software at Radialpoint. Outside of work he enjoys being a Dad, skiing, guitar, and talking about himself in third person.
Montreal 2025 sessions
OAuth, and trustworthy AI agents that don't terminate humans
English session - Beginner
There's a deluge of new AI Agents coming online, especially in the realm of customer support. LLMs are powerful, but they are non-deterministic. How do we ensure that they are trustworthy, especially as they get plugged into APIs?
I'll dismantle the myth that AI agents replace human agents, and I'll discuss the principle of end user in control in bot design. I'll also talk about a new standard proposed by the IETF that will solve this problem.
