AI Agents Podcast Building AI Agents That Work Arjun Pillai Docket AI | EP90
9 snips
Oct 23, 2025 In this engaging discussion, Arjun Pillai, co-founder of Docket.io and sales tech veteran, dives into the evolution of agentic AI for enterprise knowledge management. He explains how large language models unlock valuable unstructured data and enhance workflows. Arjun shares insights on autonomous selling, emphasizing the significance of multimodality in marketing agents, particularly through voice technologies. He also discusses the challenges of AI adoption in businesses and offers practical advice for leaders looking to transition to AI-native strategies.
AI Snips
Chapters
Transcript
Episode notes
Founder Background In Sales Tech And Exits
- Arjun built two prior companies in sales and conversational AI before Docket and exited both.
- He joined ZoomInfo, led data and account-based marketing, then left to start Docket when GPT-3.5 arrived.
Unstructured Data Holds Most Enterprise Knowledge
- Most enterprise go-to knowledge (~90%) lives in unstructured sources like calls, Slack, and people's heads.
- Large language models finally make extracting and grounding that tribal knowledge feasible at scale.
Inference-Time Compute Enabled Agentic AI
- Reasoning models introduced 'inference-time compute' that breaks tasks into iterative subtasks.
- That shift enabled agentic architectures that plan, call tools, and execute multi-step workflows.