Podcast Summary: Think AI Podcast

Episode: AI Sprawl Is Real | Priya Udeshi
Date: April 8, 2026
Host: Dave Goyal
Guest: Priya Udeshi (Chief of Staff to the CIO, Head of IT PMO at MongoDB)

Episode Overview

This episode centers on the challenges and opportunities of “AI Sprawl” within enterprises, featuring industry leader Priya Udeshi. Drawing from her extensive tech leadership at MongoDB, Priya shares how AI is being strategically woven into the fabric of enterprise IT and daily workflows, while exploring the risks, governance, and personal/professional transformations brought by rapid AI evolution. The conversation ranges from hands-on implementation stories to reflections on organizational change management, and even touches on personal experiences in using AI with children and for personal growth.

Key Discussion Points & Insights

Priya’s Background & Entry into AI

• Priya’s role: 17 years in tech, deep roots in project/program management, now Chief of Staff to MongoDB’s CIO.
• Early AI adoption: Priya highlights the evolution of AI from predictive analytics & RPA to the current era of agentic workflows.

  “AI was really a wrapper for all things predictive, analytics, machine learning, robotics, process automation… any type of automation that you could put some predictive algorithmic boundary around, was considered AI at the time.”
  — Priya [02:01]

Evolution of AI at MongoDB

• Initial Phase (RAG/Chatbot): Developed ‘Mongo GPT’—an internal tool connecting LLMs to internal data sources for secure, enterprise-specific AI interaction.
• Current Phase (Agentic workflows): Focus on creating enterprise agents for sales, HR, and democratizing agent creation for employees via low/no-code tools.

  “Enterprise scale is definitely a stream. And then also low-code, no-code, democratized agent creation is another avenue that we’re looking at.”
  — Priya [05:45]

Addressing AI Sprawl in the Enterprise

• Definition: Uncontrolled proliferation of AI tools and agents, driven by user demand and lack of central governance—parallels with “shadow IT” and SaaS sprawl.
• Risks: Security, inconsistency, and loss of control as employees build personal AI solutions outside organizational boundaries.

  “Agent sprawl is a real thing. If we don’t keep up with the pace of innovation… you’ll see people creating their own personal accounts, their own agent capabilities outside company secured environments. That’s exactly what is counter to what we want.”
  — Priya [05:25]

Navigating AI User Personas

• AI Curious: See ChatGPT as “everything.”
• AI Enthusiasts: Explore multiple models, build personal projects.
• AI Skeptics: Fear negative impacts.
• Priya identifies as an enthusiast but emphasizes humility and continuous learning.

  “Even being in the enthusiast bucket, there’s a ladder… I’m at the lower to mid rung… but the evolution of where we’ll be even a month from now will look vastly different.”
  — Priya [06:42]

Enterprise Search & Glean Agent Platform

• Challenge: Project/program teams don’t “live” in structured tools; most knowledge is in docs, emails, and unstructured sources.
• Solution: Adopted Glean as an enterprise search platform, moving away from mandates for strict data input—now exploring AI agents built on top for specialized retrieval and summarization.

  “Enterprise search… solves that problem—meet people where they are… It doesn’t try to make this one-size-fits-all.”
  — Priya [10:03]

Output vs. Outcome-Driven Leadership

• Automation and Agents: Shift focus from administrative outputs to strategic outcomes, using agents to automate routine reporting.

  “If you can automate, build agents to focus on the outputs… the shift in conversation becomes very different. You’re able to look at a data-driven report that gives you decision velocity.”
  — Priya [14:22]

Governance in the Age of AI Sprawl

• Old frameworks lag behind: Traditional intake/prioritization structures don’t keep up with AI’s rapid deployment cycles.
• New governance models needed: Lightweight, risk-balanced governance for agents—especially delineating “low-code/no-code” user tools vs. enterprise-scale applications.

  “AI governance… is not going to look the same way as traditional program governance… Just the time it takes to ask those five questions, you can spin up a solution with AI.”
  — Priya [26:18]

Responsible AI, Security, and Beta Culture

• Beta testing as norm: Beta is about informed user feedback, not stability. Must blend user and DevOps checks for AI, even in beta.

  “Beta doesn’t mean put something into production use that hasn’t been tested. Beta means—it’s beta. We’re grounded in user feedback and the user journey.”
  — Priya [30:17]

• Security & durability: Remain “table stakes,” even as AI accelerates development cycles.

Managing Hallucination & Context in AI Outputs

• Hallucination reduction: Recent models see fewer hallucinations than early LLMs. Careful prompting, embedding, and context engineering are key.

  “With certain prompt libraries… you can contain the amount of hallucinations that you get. The latest models, like with Opus… I haven’t really had that problem.”
  — Priya [33:10]

• Context as Infrastructure: Discussion of “second brains,” memory limits, context-oriented prompting, and practical tools like Obsidian and NotebookLM.

Trends & Future Directions in AI Infrastructure

• Key shifts: Memory management, prompt & context engineering, AI as infrastructure (e.g., “tiny AI” hard drives with embedded models).
• Priya’s wish: Seamless, comprehensive AI partners that preserve and leverage continuous workflow context without manual stitching.

Advice for CIOs and Tech Leaders

• Have an AI strategy—now. Not something to “wait and see.”
• Decide on architecture: Buy vs. build, point solutions, hyperscalers—must be strategic, not reactionary.
• Define governance: Lightweight frameworks that balance risk, pace of innovation, and organizational needs.

  “If you haven’t already, have an AI strategy. It’s important. This is not something to sit back and see how it evolves.”
  — Priya [42:13]

Personal Stories & Takeaways

• Empowering Kids with AI: Priya’s 7-year-old co-creates a children’s book using Gemini, images, and prompting—AI as creative amplifier, not creativity replacement.

  “There is no reason why a 7-year-old can’t enjoy the power of AI… empowering her to think beyond the bounds of what she’s able to put pen to paper on.”
  — Priya [48:29]

• AI & Intuition: Priya urges listeners (and her younger self) to trust their intuition as information density and algorithmic influence grow.

  “Your intuition… think about your gut is what’s going to tell you what your true path is and always trust that intuition.”
  — Priya [54:15]

Notable Quotes & Memorable Moments

• On AI’s Rapid Evolution:
  “The timelines around the burst of this emerging technology is so much more compressed than any other tech… With AI, even a month from now is vastly different.”
  — Priya [06:49]

• Glean’s Enterprise Search:
  “Meet people where they are… That is the power of enterprise search.”
  — Priya [10:03]

• On Hallucination:
  “At least from an end user standpoint, more hallucinations occurred with earlier models… the right reasoning, embedding, and configuration really matter.”
  — Priya [32:01]

Key Timestamps

• Priya’s AI Journey and Evolution at MongoDB – [01:29]–[05:57]
• AI User Personas & Learning Mindset – [06:39]–[08:22]
• Enterprise Search & Glean – [08:54]–[12:28]
• Agent Sprawl, Governance, and Change Management – [12:28]–[15:59]; [24:11]–[29:47]
• Beta Culture and Responsible AI – [29:47]–[31:37]
• AI Hallucination, Context Engineering, Memory – [31:37]–[41:26]
• Advice for CIOs on AI Strategy – [42:13]–[44:21]
• Measuring AI/Code Productivity – [45:23]–[46:31]
• Children’s Book AI Story – [47:39]–[50:40]
• Trusting Intuition & Advice to Younger Self – [52:57]–[54:15]

Ways to Connect & Further Learning

• Follow Priya on LinkedIn (public profile shared with the podcast)
• Find writing on Substack (Priya shares ongoing AI and tech leadership insights)
• Host Dave Goyal: Connect via Think AI Podcast, blog, and community initiatives for further discussions on AI for business and personal growth.

This episode is a must-listen for enterprise IT leaders, AI practitioners, and curious professionals navigating both promise and pitfalls of AI adoption. The real-world use cases, strategic imperatives, and “human-in-the-loop” perspective make it highly actionable for organizations and individuals alike.