20% to 30% reduction
AAR's newly launched unit Airvoyant targets a massive chunk of supply chain friction through automated procurement.
Can AI Fix Aviation's Parts Procurement Problem?
By running AI in shadow mode, airlines are testing autonomous procurement tools against real human decisions to unlock massive efficiencies.
May 26, 2026
15:29
Difficulty: Intermediate
Played
Aviation Week's MRO Podcast
Can AI Fix Aviation's Parts Procurement Problem?
By running AI in shadow mode, airlines are testing autonomous procurement tools against real human decisions to unlock massive efficiencies.
May 26, 2026
15:29
Difficulty: Intermediate
Played
TL;DR
Aviation aftermarket operations face severe supply chain bottlenecks, but agentic AI offers a path forward. This episode explores how AAR's newly launched unit, Airvoyant, uses autonomous agents to streamline parts procurement by 20% to 30% [1] — Lee Ann Shay "20% to 30% reduction: AAR's newly launched unit Airvoyant targets a massive chunk of supply chain friction through automated procurement." 00:31 . By running AI models in shadow mode, major airlines like JetBlue and Virgin Atlantic are comparing machine recommendations against historical human choices to safely train systems [3] — Jon Baker "Before letting AI run wild, systems should operate in shadow mode to compare AI recommendations against actual choices made by human procur…" 05:41 . Transitioning from basic workflows to automated decision-making allows lean procurement teams to focus on high-value, strategic efforts [2] — Jon Baker "True aftermarket transformation requires moving past simple data connections to intelligent, automated decision-making. The goal is to buil…" 00:49 .
#Aviation Aftermarket
#Supply Chain Automation
#Agentic AI
#MRO Technology
#aviation
#procurement
#artificial
#intelligence
#aftermarket
#mro
#aerospace
#supply
#chain
#automation
#efficiency
#airlines
#software
#airvoyant
#aar
Jon Baker, president and general manager of AAR's Airvoyant, joins James Pozzi and Lee Ann Shay to discuss the rollout of agentic AI in aviation parts procurement, exploring early benefits, shadow mode testing, and how lean aftermarket teams leverage automation.
Chapter list
-
James Pozzi and Lee Ann Shay welcome Jon Baker of Airvoyant to discuss using agentic AI to streamline parts procurement, outlining the unit's core goal of automating workflows and decision-making end-to-end [1] — Lee Ann Shay "20% to 30% reduction: AAR's newly launched unit Airvoyant targets a massive chunk of supply chain friction through automated procurement." 00:31 [2] — Jon Baker "End-to-end automation: The platform represents a shift from disconnected systems to a unified, autonomous agentic workforce that connects c…" 00:53 .
-
Jon Baker explains how AI delivers value in airline aftermarket operations through automated, intelligent decision-making, emphasizing the transitional role of human-in-the-loop systems that adapt based on confidence scores [1] — Jon Baker "Shadow mode benchmarking: By comparing what the system recommends with what human buyers actually chose, carriers can refine algorithm scor…" 06:07 [2] — Jon Baker "Hours of research per week: The velocity of modern AI development requires constant monitoring of new services, research papers, and softwa…" 09:08 [3] — Jon Baker "One of the things that you learn very quickly when you build these solutions is you can't build AI solutions in a vacuum." 05:44 .
-
The discussion focuses on how agents analyze data to make recommendations with clear confidence ratings, helping human teams take action while building trust toward higher levels of direct automation [1] — Jon Baker "Depressed staffing vs record demand: With teams remaining smaller and aircraft utilization running high, operators are highly receptive to …" 13:15 .
-
Jon Baker details how built-up confidence scores enable systems to automatically execute straightforward recommendations while reserving more complex choices for human review.
-
Jon Baker discusses collaborative efforts with major launch carriers, describing how historical data is utilized in shadow mode to analyze and tune the agentic workforce against historical choices[1][2].
-
The conversation covers custom-tuning AI tools to match individual carrier needs and leveraging early feedback to explore upstream functionality like requisition demand optimization[1].
-
Jon Baker addresses the high velocity of modern AI development, data ownership, security, and the industry's adoption drive fueled by reduced post-COVID staffing and ongoing demand pressures[1][2][3].
- Agentic AI
- A class of artificial intelligence systems capable of autonomous planning, decision-making, and executing multi-step workflows to achieve specific goals.
- Human in the Loop
- A design pattern in automation where an artificial intelligence system provides analysis and recommendations, but a human must review and authorize the final action.
- Shadow Mode
- A testing methodology where a new software system or AI model runs in parallel with active operations, receiving real production data to generate mock recommendations without affecting live workflows.
- Data Lake
- An environment or repository where vast amounts of raw data are stored in their native format until needed for analytics or machine learning applications.
- Digital Twin
- A virtual or digital representation of a physical object, process, or system used to simulate behavior and analyze performance under varying conditions.
Chapter 1 · 00:00
Introducing Airvoyant and Agentic AI
James Pozzi and Lee Ann Shay welcome Jon Baker of Airvoyant to discuss using agentic AI to streamline parts procurement, outlining the unit's core goal of automating workflows and decision-making end-to-end [1] — Lee Ann Shay "20% to 30% reduction: AAR's newly launched unit Airvoyant targets a massive chunk of supply chain friction through automated procurement." 00:31 [2] — Jon Baker "End-to-end automation: The platform represents a shift from disconnected systems to a unified, autonomous agentic workforce that connects c…" 00:53 .
The Real Goal of Agentic AI in MRO
True aftermarket transformation requires moving past simple data connections to intelligent, automated decision-making. The goal is to build an agentic workforce that manages transactions and systems end-to-end.
End-to-end automation
The platform represents a shift from disconnected systems to a unified, autonomous agentic workforce that connects customers and vendors.
Chapter 2 · 02:15
Delivering Value via Intelligent Decisions
Jon Baker explains how AI delivers value in airline aftermarket operations through automated, intelligent decision-making, emphasizing the transitional role of human-in-the-loop systems that adapt based on confidence scores [1] — Jon Baker "Shadow mode benchmarking: By comparing what the system recommends with what human buyers actually chose, carriers can refine algorithm scor…" 06:07 [2] — Jon Baker "Hours of research per week: The velocity of modern AI development requires constant monitoring of new services, research papers, and softwa…" 09:08 [3] — Jon Baker "One of the things that you learn very quickly when you build these solutions is you can't build AI solutions in a vacuum." 05:44 .
Leveraging AI Confidence Scores
AI agents build trust by assigning confidence scores to their recommendations based on available data and timing. As these scores improve, systems can safely transition from human review to fully autonomous execution.
Chapter 3 · 05:26
Confidence Scores and Human-in-the-Loop
The discussion focuses on how agents analyze data to make recommendations with clear confidence ratings, helping human teams take action while building trust toward higher levels of direct automation [1] — Jon Baker "Depressed staffing vs record demand: With teams remaining smaller and aircraft utilization running high, operators are highly receptive to …" 13:15 .
The Power of Shadow Mode Testing
Before letting AI run wild, systems should operate in shadow mode to compare AI recommendations against actual choices made by human procurement teams. This provides a risk-free environment to calibrate and tune performance.
Shadow mode benchmarking
By comparing what the system recommends with what human buyers actually chose, carriers can refine algorithm scoring without risk.
Chapter 4 · 07:49
Transitioning to Direct Automation
Jon Baker details how built-up confidence scores enable systems to automatically execute straightforward recommendations while reserving more complex choices for human review.
Surviving the Velocity of AI Development
The sheer speed of AI innovation is unprecedented, requiring software teams to spend hours each week studying new research. Organizations must use AI-assisted development tools just to build and scale their own AI agent systems.
Chapter 5 · 08:49
Collaboration with Launch Partners
Jon Baker discusses collaborative efforts with major launch carriers, describing how historical data is utilized in shadow mode to analyze and tune the agentic workforce against historical choices[1][2].
Hours of research per week
The velocity of modern AI development requires constant monitoring of new services, research papers, and software design patterns.
Monetizing Aggregated Aftermarket Data
While individual airlines struggle to monetize isolated history, aggregating anonymized procurement data unlocks powerful trends. Letting agentic systems act on these collective insights is where the ultimate value lies.
Chapter 6 · 10:35
Custom Tuning and Future Functionality
The conversation covers custom-tuning AI tools to match individual carrier needs and leveraging early feedback to explore upstream functionality like requisition demand optimization[1].
Post-COVID Lean Teams Turn to AI
With procurement and aftermarket staffing levels still depressed compared to pre-COVID times and travel demand back at record highs, airlines are aggressively turning to AI to bridge the operational gap.
Chapter 7 · 12:49
AI Development Velocity and Industry Adoption
Jon Baker addresses the high velocity of modern AI development, data ownership, security, and the industry's adoption drive fueled by reduced post-COVID staffing and ongoing demand pressures[1][2][3].
Depressed staffing vs record demand
With teams remaining smaller and aircraft utilization running high, operators are highly receptive to efficiency-driving AI tools.
No indexed bits in this chapter.
Show stoppers
The Real Goal of Agentic AI in MRO
True aftermarket transformation requires moving past simple data connections to intelligent, automated decision-making. The goal is to build an agentic workforce that manages transactions and systems end-to-end.
The Power of Shadow Mode Testing
Before letting AI run wild, systems should operate in shadow mode to compare AI recommendations against actual choices made by human procurement teams. This provides a risk-free environment to calibrate and tune performance.
Depressed staffing vs record demand
With teams remaining smaller and aircraft utilization running high, operators are highly receptive to efficiency-driving AI tools.
Snapshots ()
Key Quotes ()
This episode
Claims & Sources
2 / 3 cited (67%)
Factual claims made this episode, and whether a source was named.
Airvoyant was launched in April 2026 to streamline parts procurement in aviation using agentic AI.
Airvoyant's technology is designed to streamline parts purchases for airlines and MROs by 20% to 30%.
Procurement and aftermarket teams generally have not recovered to their pre-COVID staffing levels.
This episode
Cast
-
AAR's newly launched business unit focused on agentic AI parts procurement.
-
A major global aviation services provider and parent company of Airvoyant.
-
A major US carrier collaborating with Airvoyant as a launch partner.
-
A prominent UK airline collaborating with Airvoyant as a launch partner.
-
The flag carrier of Canada and one of Airvoyant's launch partners.
-
One of the global airlines named as a launch partner for Airvoyant.
-
The national carrier of Thailand and one of Airvoyant's global launch partners.
-
An MRO software suite under the AAR umbrella mentioned alongside aftermarket use cases.
Stats
Sponsors
Related episodes
No links parsed
We scan show notes for social handles, websites and apps. Nothing matched on this episode.
