Aviation safety depends on having the right information at the right moment. The problem is that the information is fragmented, voluminous, and hard to retrieve when no flight is entirely the same. In non-standard situations, crews aren’t short on rules—they’re short on time to find and verify the one that matters.

Leo Kotil built Overwatch AI after spending a decade in the cockpit and airline operations. His bet is simple: GenAI is most valuable in aviation when it turns manuals, NOTAMs, and operational data into fast, source-backed answers that crews can verify—and still use when connectivity drops.

This conversation examines where AI belongs in flight operations. Not in decision authority, but in compressing the time from a non-standard situation to the verified reference that governs it without breaking the safety model that makes aviation work. 

What You’ll Learn

• Why aviation’s real bottleneck is retrieval, not knowledge: The documents exist; the problem is locating the right section fast enough when the situation isn’t standard.
• How “digital” still leaves crews doing manual search work: iPads and PDFs replaced paper, but many workflows still rely on folder navigation and keyword search.
• What makes an AI assistant usable in regulated ops: Answers must surface the exact source passages so pilots and frontline teams can confirm and trust the output.
• Why context is the product, not a nice add-on: Pulling in live and structured data (weather, aeronautical publications, flight context) removes extra steps and reduces mistakes.
• How multilingual reality changes system design: Crews ask in their native language while documents stay in English, often mixing aviation terms—retrieval has to handle that reliably.
• How startups ship into aviation within a regulated environment: Deploy as a supplemental layer on top of existing certified tools, then prove value before becoming “core.”
• Why offline capability is mandatory: Aviation software needs a usable fallback when connectivity is unavailable, not just a degraded mode in theory.
• The tradeoffs of using proprietary LLM APIs in airlines: Provider dependency, infrastructure variability, and sensitive data processing create risks beyond normal cloud hosting.

Time-Stamped Highlights

• (00:20) From Airline Pilot to Aviation Founder
• (01:08) Career Path Across Airline and Business Aviation
• (03:01) The Operational Reality of a Pilot’s Day
• (05:54) Pre-Flight Procedures and Checklist Pressure
• (09:10) EFBs, Manuals, and Information Overload
• (12:12) Founding Overwatch AI and Meeting a Co-Founder
• (13:56) Early Traction and the Techstars Accelerator
• (15:41) Networks, Credibility, and Selling Into Airlines
• (20:04) Designing an AI Assistant for Frontline Ops
• (21:29) Native Language, Voice Input, and Real Usage
• (23:01) Regulatory Constraints and Compliance Strategy
• (27:01) Product Roadmap and Near-Term Focus
• (29:09) Contextual Data as the Core Differentiator
• (37:39) Offline AI, Edge Constraints, and Aviation Grade
• (40:52) Model Lock-In, APIs, and Enterprise Risk Tradeoffs 

Guest

Leo Kotil — Founder and CEO, Overwatch AI
Leo is the co-founder and CEO of Overwatch AI, where he is building AI systems to support pilots, cabin crew, operations control centers (OCC), and ground staff in managing flight disruptions and non-standard operations. A former airline pilot, he brings firsthand experience of aviation frontline workflows and operational decision-making into the design of practical, regulation-aware AI tools for airlines.
LinkedIn: https://www.linkedin.com/in/leo-kotil/
Company: https://overwatch-ai.com/

About the Podcast

Travel Tech Podcast features long-form conversations with leaders across travel and technology. The show explores how software, data, operations, and distribution come together in real businesses, with an emphasis on tradeoffs, incentives, and lessons that transfer beyond any single company or role.

Host

Alex Brooker — Founder, Airside Labs
Alex is an engineer, technology leader, and founder with deep expertise in mission-critical systems and AI oversight. He leads Airside Labs, an AI business that applies aviation-grade testing and compliance rigor to enterprise AI systems, helping organizations build and test AI agents in regulated environments. Before founding Airside Labs, Alex built and scaled complex software in aviation and safety-critical domains, blending product innovation with disciplined engineering practices. He also invests in early-stage technology ventures and advocates for thoughtful, real-world AI deployment strategies.
LinkedIn: https://www.linkedin.com/in/alex-brooker-2280002/

Links & References

• EASA (European Union Aviation Safety Agency): https://www.easa.europa.eu/
• Electronic Flight Bag (EFB): https://www.easa.europa.eu/en/domains/operations/electronic-flight-bag-efb
• NOTAMs (Notices to Air Missions): https://www.icao.int/airnavigation/information-management/notams/Pages/default.aspx

Brought To You By

Airside Labs — Airside Labs supports aviation and travel operators with tools to test, deploy, and scale modern data and AI systems in safety-critical environments. Learn more at https://airsidelabs.com.

Most drone use cases fail for a surprisingly mundane reason: they can’t safely or legally scale past a few hundred meters. The aircraft are capable of flying kilometers, but operations collapse once you factor in regulatory limits, detection physics, and fragile surveillance infrastructure.

James Dunthorne has encountered this constraint from every angle. From PhD research on collision avoidance to early agricultural drone deployments and high-precision surveying over railways and landmark sites, he’s seen exactly where theory breaks when exposed to real airspace.

This conversation digs into what actually blocks BVLOS operations: mixed transponder environments, latency requirements measured in seconds, why centralized flight-tracking systems struggle under regulatory scrutiny, and how edge-based sensor networks change what’s possible for drones, aviation, and AI-driven systems in the physical world.

What You’ll Learn

• Why BVLOS is an infrastructure problem, not a drone problem: Aircraft capability is rarely the constraint; reliable surveillance and separation in mixed low-altitude airspace is.
• Why visual line of sight is a weak safety mechanism: Human vision and standard cameras fail at meaningful ranges given aircraft closing speeds.
• How electronic conspicuity fragments the airspace picture: Multiple non-interoperable transponders and non-transponding aircraft create unavoidable gaps.
• Why multilateration enables detection without GPS: Legacy Mode S aircraft can be located using precise timing across multiple ground sensors.
• Why centralized flight-tracking systems fail safety-critical tests: Single points of failure, variable latency, and opaque architectures undermine regulatory confidence.
• How edge networks localize failure and reduce latency: Direct sensor-to-consumer connections keep surveillance resilient and deterministic.
• Why incentives matter when scaling physical networks: Revenue sharing turns infrastructure deployment into a distributed, maintainable system.
• How this architecture extends beyond aviation: A real-time physical-world data layer becomes foundational for AI, robotics, and autonomous systems.

Time-Stamped Highlights

• (01:10) Aerospace Engineering and Autonomous Systems Origins
• (03:10) Collision Avoidance and Regulatory Safety Limits
• (05:10) Agricultural Drones and NDVI in Practice
• (07:10) The 500-Meter Constraint and Operational Inefficiency
• (09:15) From Drones to GIS and Surveying
• (12:10) High-Accuracy Rail Inspections From the Air
• (14:30) Discovering the BVLOS Bottleneck
• (18:10) Why Closing Airspace Doesn’t Scale
• (22:00) Radar, Cameras, and Detection Physics
• (26:15) Transponder Fragmentation in Low Airspace
• (30:15) Multilateration and Time Synchronization
• (34:20) Why Surveillance Must Be Ground-Based
• (37:40) Latency, Reliability, and Centralized SaaS Limits
• (41:30) Edge Networks and Failure Localization
• (46:10) Cyber Risk in Safety-Critical Systems
• (51:05) Scaling Sensors With Economic Incentives
• (54:30) From Local Coverage to Global Networks
• (58:10) AI, Software Moats, and Physical Data
• (01:03:40) Travel, Trust, and the Return of Face-to-Face

Guest

James Dunthorne — CEO & Co-Founder, Neuron
James is the CEO and co-founder of Neuron, where he leads the development of edge-based sensor networks and surveillance infrastructure that enable beyond-visual-line-of-sight drone operations in mixed airspace. He brings over 15 years of experience across aerospace engineering, autonomous systems, drone operations, and high-accuracy surveying, with a background spanning academic research, regulated aviation environments, and real-world deployments.
LinkedIn: https://www.linkedin.com/in/jamesdunthorne/

About the Podcast

Travel Tech Podcast features long-form conversations with leaders across travel and technology. The show explores how software, data, operations, and distribution come together in real businesses, with an emphasis on tradeoffs, incentives, and lessons that transfer beyond any single company or role.

Host

Alex Brooker — Founder, Airside Labs
Alex is an engineer, technology leader, and founder with deep expertise in mission-critical systems and AI oversight. He leads Airside Labs, a consultancy that applies aviation-grade testing and compliance rigor to enterprise AI systems, helping organizations uncover bias, privacy risks, and governance gaps in regulated environments. Before founding Airside Labs, Alex built and scaled complex software in aviation and safety-critical domains, blending product innovation with disciplined engineering practices. He also invests in early-stage technology ventures and advocates for thoughtful, real-world AI deployment strategies.
LinkedIn: https://www.linkedin.com/in/alex-brooker-2280002/

Links & References

• Neuron: https://neuron.world
• Neuron on X: https://x.com/neuron_world
• Jet Vision (flight surveillance partner referenced)
• NDVI (Normalized Difference Vegetation Index)
• ADS-B, Mode S, and multilateration (aviation surveillance concepts)
• 4D Sky: https://4dsky.com
• Land’s End Airport (operational deployment referenced)
• NATO (project referenced)
• Gatwick Airport drone incident (airspace security reference)

Brought To You By

Airside Labs — Airside Labs supports aviation and travel operators with tools to test, deploy, and scale modern data and AI systems in safety-critical environments. Learn more at https://airsidelabs.com.

AI is making knowledge work faster — but it’s also surfacing an uncomfortable tension: when the “doing” becomes cheap, the limiting factor shifts to everything humans do around it. This tension shows up in two places at once: inside engineering teams (identity, craft, and maintainability) and inside go-to-market (trust, distribution, and buying behavior).

In this episode, Ian Painter, Oliver Deakin, and Adrian McKenzie approach this from lived experience rather than speculation. They have built and scaled data-intensive travel technology, operated deep inside enterprise environments, and navigated acquisition into a public-market business. Instead of defaulting to debates about job loss, they focus on a more operational problem: when building no longer creates advantage on its own, what does?

What You’ll Learn

• Why speed shifts the bottleneck rather than removing it: As AI compresses build cycles, advantage moves from execution to decision-making, positioning, and trust.
• How identity shapes resistance to AI tools: Engineers most attached to craft and code quality often struggle more than those focused on outcomes.
• Why “good enough” AI output is still valuable: Treating AI like a junior teammate reframes imperfection as leverage rather than failure.
• Where maintainability breaks in mixed human-AI teams: Code that functions can still create long-term friction when humans need to read, test, and evolve it.
• How startup time-to-market dynamics are collapsing: Mockups, demos, and customer conversations now happen days into company formation.
• Why distribution may matter more than differentiation: When demos converge, embedded relationships and brand trust regain power.
• How build-versus-buy decisions may flip: Internal teams coordinating many agents could replace procurement with custom internal builds.
• Why data becomes the defensible asset again: As software commoditizes, curated, hard-earned datasets grow in relative value.
• What near-term “seniority” may look like: Capability may increasingly be measured by how many agents someone can effectively coordinate.
• How to prepare students for knowledge work amid AI: First-principles thinking, critical evaluation, and tool fluency matter more than any single technology.

Time-Stamped Highlights

• (00:32) AI, Jevons Paradox, and the Framing Question
• (01:37) AI Acceleration in Knowledge Work
• (02:19) Ian Painter’s Founder Perspective
• (03:19) Oliver Deakin on Modern Engineering Practice
• (03:42) Adrian McKenzie on Leadership and Teams
• (05:24) Engineers’ Emotional Responses to AI
• (07:05) Why Imperfect AI Gets Dismissed
• (08:26) Hands-On Experience With AI Coding Tools
• (09:51) Functional Code Versus Maintainable Systems
• (11:26) Startup Dynamics in an AI-Accelerated World
• (13:07) Speed to Market and Competitive Compression
• (15:05) Sales, Marketing, and Distribution Shifts
• (19:42) Humans as the Limiting Factor
• (22:00) Brand Trust and Embedded Distribution
• (35:03) Data as the Enduring Moat
• (42:15) Advice for Future Knowledge Workers

Guests

Ian Painter — Startup Advisor and Mentor. Previously, Vice President, Platform and Data at Cirium; Founder, Snowflake Software
Ian is a seasoned technology leader in aviation data and analytics. He founded Snowflake Software in 2001, building enterprise data exchange and aviation data platforms that were later acquired by Cirium (RELX plc). As VP of Platform and Data, he oversaw data strategy and large-scale platform initiatives at one of the world’s most trusted aviation analytics companies.
LinkedIn: https://www.linkedin.com/in/ianpainter/

Oliver Deakin — Fractional CTO, Advisor and previously Technology Leader at Cirium, Former Snowflake Software CTO, and Senior Engineer at IBM
Oliver has served in senior technical leadership roles, including as CTO at Snowflake Software during its rise in aviation data solutions. He has deep practical experience with software architecture, developer tooling, and emerging technologies applied to complex domains like travel and real-time data systems.
LinkedIn: https://www.linkedin.com/in/olideakin/

Adrian McKenzie — Director of Software Engineering at Cirium
Adrian leads engineering teams responsible for delivering scalable, mission-critical aviation data and analytics solutions. His background includes progressive leadership in software delivery and architecture at both Snowflake Software and Cirium, with decades of experience in team performance, engineering operations, and large-scale systems.
LinkedIn: https://www.linkedin.com/in/adrianmckenzie/

About the Podcast

Travel Tech Podcast features long-form conversations with leaders across travel and technology. The show explores how software, data, operations, and distribution come together in real businesses, with an emphasis on tradeoffs, incentives, and lessons that transfer beyond any single company or role.

Host

Alex Brooker — Founder, Airside Labs
Alex is an engineer, technology leader, and founder with deep expertise in mission-critical systems and AI oversight. He leads Airside Labs, a consultancy that applies aviation-grade testing and compliance rigor to enterprise AI systems, helping organizations uncover bias, privacy risks, and governance gaps in regulated environments. Before founding Airside Labs, Alex built and scaled complex software in aviation and safety-critical domains, blending product innovation with disciplined engineering practices. He also invests in early-stage technology ventures and advocates for thoughtful, real-world AI deployment strategies.
LinkedIn: https://www.linkedin.com/in/alex-brooker-2280002/

Links & References

• Jevons Paradox and efficiency-driven demand
• AI tools mentioned: GitHub Copilot, Claude
• Concepts discussed: software commoditization, distribution moats, curated data assets, agent-based development, human-in-the-loop systems

Brought To You By

Airside Labs — Airside Labs helps organizations deploy AI safely and responsibly by applying aviation-grade testing, assurance, and oversight to complex systems. Learn more at https://airsidelabs.com