In this episode of AI Leadership Lab, host Ryan Heath sits down with Philip Rathle, Chief Technology Officer at Neo4j, to explore how graph databases are revolutionizing AI infrastructure and enterprise knowledge systems.

Philip reveals why understanding the relationships between data points is more powerful than having all the facts, and how companies like Google built trillion-dollar businesses on graph algorithms. From explaining knowledge graphs in plain language to discussing how graph-based retrieval can make AI more trustworthy and explainable, this conversation delivers actionable insights for leaders seeking to build more effective AI systems.

Takeaways

Relationships Matter More Than Facts

Understanding connections between data points often reveals more than the data itself. Philip demonstrates this with a striking example: knowing how friends-of-friends-of-friends behave is a better predictor of someone's behavior than having comprehensive facts about that individual person. This principle applies across business contexts, from customer 360 systems to organizational analysis.

The Real vs. Declared Org Chart

Graph technology can reveal an organization's true power structure by analyzing email patterns, Slack messages, and information flows. Companies are using this to identify single points of failure—like one person receiving all questions on a critical topic—and to facilitate warm introductions by mapping who knows whom across company boundaries.

Graph RAG Delivers Better Results with Less

By combining knowledge graphs with language models, companies are achieving superior answers while using two-thirds less data in context windows. This "graph RAG" approach queries a knowledge graph first, then feeds only the most relevant results to the model, resulting in faster responses, lower costs, and reduced energy consumption.

AI Systems Need Knowledge Layers, Not Just Language Models

Language models alone have fatal flaws for enterprise use: they hallucinate, lack company-specific data, operate as black boxes, and can't discern what information is appropriate for which purpose. Successful AI implementations complement LLMs with knowledge graphs that provide exact, explainable results while maintaining the context and causality that business users understand.

Explainability is the Path to Trust and Adoption

Graph-based systems enable accountability by providing traceable answers.

Timestamps

[00:00] Introduction

[01:12] Philip's journey from consulting to graph databases

[04:00] Facebook and Google as graph pioneers

[05:18] What is a knowledge graph?

[07:44] The true org chart: mapping real power structures

[09:30] Making AI more explainable and trustworthy

[14:13] Build vs. buy considerations for graph technology

[16:07] How graphs will reshape AI infrastructure

[18:08] Graph RAG and the future of AI applications

[20:00] Human impact: accountability and agency in AI

About the Guest

Philip Rathle is the Chief Technology Officer at Neo4j, a company that has been pioneering graph database technology and knowledge graphs for AI applications. Philip's career began in consulting, where he quickly became convinced that data serves as a mirror of business operations — the better your data, the better handle you have on your business. He built United Airlines' first passenger 360 system.

Connect with Philip & Neo4j

Neo4j Website: https://neo4j.com

LinkedIn: Search for Philip Rathle, CTO at Neo4j

Support the Show

If you'd like to appear on the show or know someone who should be featured, visit RyanHeathConsulting.com. Please leave a five-star rating or review to help more leaders discover these insights.

Episode Overview

In this episode of AI Leadership Lab, host Ryan Heath speaks with Pari Parchi, Founder and CEO of Panorama Aero, about the critical infrastructure challenges facing America's airspace.

With the US still operating on World War II-era radar systems while drones proliferate and autonomous flight technology advances, Pari reveals where the private sector may need to take more airspace management into its own hands. From the regulatory gridlock preventing counter-drone technology to the looming pilot shortage forcing autonomous solutions, this conversation exposes the urgent tensions between technological capability and outdated oversight systems.

Key Takeaways

America's Airspace Runs on World War II Technology

U.S. airspace management still relies on infrastructure dating to World War II, with radar systems and radio control as the foundation. Most aircraft landings remain VFR (visual flight rules), meaning pilots land by sight rather than automated systems. Since the 2003 ATC NextGen bill aimed at modernization, only 16% of initiatives have been completed.

The Drone Regulation Paradox

If someone flies a drone into your backyard to look through your windows, shooting it down is illegal — but the drone operator usually faces no penalty. This regulatory gap, primarily under Federal Communications Commission jurisdiction, leaves Americans vulnerable to privacy violations and potential security threats. The U.S. is up to two years behind Ukraine, Israel, and China in drone and counter-drone technology development, partly because we're not dealing with these threats daily.

The Private Sector Will Lead Airspace Security

With federal agencies stretched thin and regulatory changes moving slowly, private sector organizations are developing their own airspace protection systems. Companies are deploying counter-drone sensors to protect critical infrastructure, airports, public events, and private property. While they may not be able to shoot down unauthorized drones, they can identify operators, track license plates, and locate individuals for enforcement action.

The Pilot Shortage Will Force Autonomous Flight

At $1,000 to $1,500 per day, human pilot costs for the smallest aircraft can be economically infeasible: think four- or six-seater eVTOL vehicles and flying cars. The global pilot shortage is therefore increasingly the inevitability of autonomous flight. The transition will likely start with reducing commercial aircraft from two pilots to one, with AI serving as a "backseat driver" co-pilot.

Humans and Machines See the Airspace Differently

While AI can handle routine flight paths, human pilots provide irreplaceable value during emergencies, mechanical failures, and unexpected weather conditions. Having physical presence in the aircraft versus ground-based command and control is like attending the Super Bowl in person versus watching on TV.

Special Mission Aircraft Protect More Than We Realize

Turboprop aircraft and business jets serve critical public safety functions: surveillance, reconnaissance, mapping, medevac, and firefighting. These "special mission" or "multi-mission" aircraft use the airframe as a technology chassis, implementing specialized equipment for essential operations. The complexity and cost of maintaining these assets is widely underestimated.

About the Guest

Pari Parchi and Panorama Aero specialize in the acquisition and management of specialized aerospace assets. Through defense, aerospace, and early-stage investing experience, Pari brings a unique global perspective to airspace management challenges, having lived and worked across four continents.

Panorama Aero focuses on special mission and multi-mission aircraft — turboprop aircraft and business jets modified for specific purposes including surveillance, reconnaissance, mapping, medevac, firefighting, and other critical operations.

LinkedIn: linkedin.com/in/pariparchi

Company: panorama.aero

Episode Overview

In this episode of AI Leadership Lab, host Ryan Heath sits down with Ryan Steelberg, CEO of Veritone, to explore the practical realities of deploying AI in enterprises. With a deep history in ad tech and in structuring previously unstructured audio and video data, Steelberg offers a grounded perspective on AI adoption that cuts through the hype. From discussing the critical importance of data infrastructure to sharing insights on ROI measurement and the mistakes companies make when integrating AI, this conversation provides essential guidance for leaders who want AI solutions that actually work—not just shiny marketing promises.

Key Takeaways

Focus Data Infrastructure, Forget AI Magic

Most organizations struggle with basic data management and cloud migration before they can meaningfully apply AI. Companies must understand and embrace their data journey first—there's no skipping this step, regardless of how advanced the AI tools promise to be.

AI is a Tool, Not a Solution

When evaluating AI products, redact every mention of "AI" from the marketing literature and ask: why are you buying this software? The AI is just a component, like an engine in a car. Focus on whether the solution satisfies your well-defined needs, not whether it's labeled as "next generation" or "future proof."

Track Everything to Improve Everything

Smart AI deployment requires comprehensive tracking of how users interact with applications. This data reveals whether bottlenecks stem from the AI model itself or the application layer, enabling companies to improve both the technology and the workflow continuously.

Customized ROI Metrics Matter

ROI metrics must be tailored to specific use cases and business models. What drives value for a sports organization (speed to market for content) differs radically from what matters to a media company (ad revenue optimization), even when using the same technology stack.

Combine Experience with Fresh Perspective

Organizations need both veterans who understand traditional processes and newcomers who organically embrace AI tools, and communicate naturally with data.

Regulated Environments Require Specific AI Approaches

In secure or air-gapped environments like Department of Defense networks, you cannot invoke third-party AI models. Everything must be containerized and deployable within the secure environment.

Key Quotes

"Imagine taking a piece of marketing literature and redacting any word that mentions AI. Why are you buying this software solution?"

"Don't ever throw away your ore. You don't know where the gold or diamonds are gonna be materialized or processed through."

Chapter Timestamps

[00:00] Veritone's AI journey from ad tech origins

[02:04] Bringing structure to unstructured data

[04:02] Deploying AI in regulated industries

[05:17] Product roadmap evolution and customer feedback

[08:00] Common mistakes in AI integration

[10:06] Skills and upskilling challenges

[12:25] Measuring ROI in AI deployments

[16:00] Surprising customer use cases

[21:00] Smart questions for evaluating AI products

About the Guest

Ryan Steelberg is the CEO of Veritone. Steelberg's journey into AI began with a fundamental problem: how to target ads against audio and video content in an increasingly organic media ecosystem. This challenge led Veritone to develop sophisticated capabilities in transcription, object detection, and machine vision to bring structure to unstructured media content.

Under Steelberg's leadership, Veritone's major clients include NBCUniversal, iHeartMedia, the US Tennis Association, CNBC, and the Department of Defense.

Connect with Ryan & Veritone

https://www.veritone.com

https://linkedin.com/in/ryansteelberg/

About AI Leadership Lab

AI Leadership Lab interviews C-suite leaders about their AI journeys, covering what's working, where they're getting stuck, how they pivot, and the lessons they take from the losses.

Host: RyanHeathConsulting.com