From Hype To ROI: Making Enterprise AI Work
Manage episode 516346989 series 3499431
Interested in being a guest? Email us at [email protected]
Most AI pilots impress in a demo and stall in production. We wanted to unpack why. Joseph Kim is the President and CEO of Druid AI joins us to break down the real blockers to AI readiness and the playbook that moves teams from wow to working systems. We talk candidly about market hype, sobering stats on failed deployments, and the ingredients that make enterprise agents safe, accurate, and compliant at scale.
We start with first principles: define the business case, choose the right capability for the job, and decide if AI is even necessary. Joe draws a clean line between generative tasks and agentic tasks, explaining why a passable answer is fine for search but dangerous for automated actions. He shows how Druid engineers fidelity with background subroutines that pre-check identity, history, and likely intent, so the first response is accurate and the downstream steps are auditable. That control layer—complete logs, policy dials, and compliance hooks—turns black-box models into systems you can govern and certify.
From there, we tackle data hygiene, hallucination traps, and how to measure success beyond vanity metrics. Joe lays out a practical path: start small with a high-value workflow, track NPS, deflection rates, and accuracy, then expand into adjacent use cases. For enterprises running dozens of agents from different vendors, we dig into orchestration. Druid’s Conductor acts as a control plane that routes tasks, mediates context, and prevents lock-in, so you can swap components without burning time and budget. We close with a grounded take on the AI bubble: yes, excitement inflates expectations, but the durable gains will mirror the early internet—messy at first, transformative over time.
If you’re ready to turn experiments into outcomes, this conversation will help you align technology with ROI, design for compliance, and scale with confidence. Subscribe, share with a teammate who owns an AI pilot, and leave a review with the one blocker you want us to tackle next.
More at https://linktr.ee/EvanKirstel
Chapters
1. From Hype To ROI: Making Enterprise AI Work (00:00:00)
2. Setting The Stage: AI Readiness (00:00:01)
3. Joe’s Background And Druid’s Mission (00:00:26)
4. Market Hype Versus Production Reality (00:01:03)
5. Accuracy And Control Beyond LLMs (00:02:17)
6. Why 87 Percent Aren’t Ready (00:04:10)
7. Assessing Readiness And ROI (00:05:47)
8. Generative Versus Agentic Use Cases (00:06:43)
9. From Wow To Work: Druid’s Approach (00:07:34)
10. Misconceptions, Failures, And Lessons (00:08:41)
11. Orchestrating Many Agents With Conductor (00:10:20)
12. Are We In An AI Bubble (00:12:15)
13. Events, Next Steps, And How To Connect (00:13:36)
14. Closing And Media Plug (00:15:58)
542 episodes