Summary
Kartik described enterprise AI agents moving from internal demos into customer care, coding, network automation, and internal workflows like RFPs and finance forecasting. In customer service, the company is replacing IVR with AI, expanding from question-answering to safe actions, while using evals, guardrails, and partners like Decagon to scale without locking into one solution.
Key Topics
5 key topics from Ashwin Sreenivas, Kartik Sheth and Hugh Langley at HumanX Amsterdam 2026. Thicker branches were mentioned more often in the talk.
Past the Demo: Deploying AI Agents at Enterprise Scale
Click a topic to open it
Key Takeaways
- Start with safe, high-volume actions before expanding agent autonomy.
- Build evals and guardrails into the product from the beginning.
- Use partners to accelerate deployment while owning core harnesses.
- Expect containment rates to climb gradually, with edge cases hardest.
- Design adapters and caching for legacy systems not built for agents.
“We've gone from features as the primary currency to evals as a primary currency and guardrails.”
Share
Up Next
DJL
WatchNext in agenda
Rewriting the Operating Model for the AI Era
Des Traynor, Jennifer Smith, Luisa Bomke
ai agentsenterprise ai
CN
WatchAlso on ai agents
Is AI Killing Software as a Service?
Clay Bavor, Niall Firth
ai agentssaas
TMH
WatchAlso on ai agents
Is This the "End of SaaS" Moment We’ve Been Building For?
Tom Eggemeier, Mike Murchison, Hugh Langley
saas pricingai agents