Ep157: Beyond the Hype: Real-World AI Agent Deployments at Automation Anywhere, DataVisor, and Sumo Logic

Panel brings together ISV leaders from automation, fraud detection, and security operations.

Companies rethinking entire business processes rather than automating incremental portions with agents.

Start with immutable data before tackling real-time changing data in production.

Intent for change must come from board, CEO, and customers simultaneously.

Challenge: proving agent value beyond CSAT when internal teams block deployment.

Sumo Logic measures Mean Time to Resolution, aiming to cut hours to zero.

DataVisor cuts fraud alert resolution from one hour down to twenty minutes.

Customers demand reliability as workflows shift from deterministic to probabilistic agent decisions.

Automation Anywhere spent three years making every platform component fully agent-ready.

Focus on business outcomes, not chasing every new model release each week.

Human oversight still critical—agents are task-oriented and prone to hallucinations and drift.

Humans validate agent findings, then let agents scale actions across hundreds instances.

Pricing experiments range from platform-plus-consumption to outcome-based to decision-event models.

Token pricing doesn't work due to varied data modalities and complexity.

Next two quarters: more POCs moving to production with productive agents deployed.

Future prediction: enterprise apps becoming systems of knowledge powered by MCP protocol.