Airside Labs
    Analysis

    Top Aviation AI Startups Building the Future of Flight in 2026

    •Updated •Airside Labs Team
    Top Aviation AI Startups Building the Future of Flight in 2026

    An honest, sourced map of the startups building aviation AI in 2026 — flight operations, cockpit copilots, autonomy, and the truth about aviation foundation models — plus every travel and aviation tech company whose founder has been on the Travel Tech Podcast.

    Top Aviation AI Startups Building the Future of Flight in 2026

    Aviation AI has moved from conference slideware to signed contracts. In the last year, an AI flight-operations company won a multi-hundred-million-dollar FAA programme, an AI copilot startup picked up a US special-operations contract, and an autonomy firm crossed a $12 billion valuation. If you are trying to understand who is actually building aviation AI — and, just as importantly, how anyone knows these systems are safe — here is an honest, sourced map of the field.

    A note on scope: "aviation AI" is not one market. A route-optimisation model, a vision system that spots traffic out the window, and an autopilot that lands a Cessna are completely different engineering problems. We have grouped the companies by what they actually build, flagged where the evidence is strong versus thin, and resisted the temptation to pad the list with firms that only use AI at the edges. Market-size figures in this space vary wildly between research houses; the most conservative widely-cited estimate puts AI in aviation at roughly $7.45 billion in 2025, rising to about $8.83 billion in 2026, with predictive maintenance the single largest application (Fortune Business Insights). Treat that as a vendor estimate, not gospel — others range from under $2 billion to over $8 billion for the same year.

    The second half of this guide is different in kind. It covers every company whose founder or leader has sat down with us on the Travel Tech Podcast, from drone traffic management to hotel data. Those entries come from the conversations themselves, and each one links to the episode so you can hear the founder explain the business in their own words.

    The Field at a Glance

    CompanySegmentWhat they buildNotable marker
    Air Space IntelligenceFlight ops / ATMML route & flight-operations optimisation (Flyways)Won the FAA's ~$875M, 12-year "SMART" ATM programme (2026)
    Beacon AICockpit copilot"Murdock" AI copilot + "Lighthouse" data platform$15M Series A; $49.5M US SOCOM contract (2026)
    Shield AIAutonomy (defence)"Hivemind" autonomous AI pilot~$12.7B valuation after a 2026 raise
    Merlin LabsAutonomyAI autopilot software for existing aircraftWent public for ~$200M (2026)
    DaedaleanPerceptionML / computer-vision autonomous-flight avionicsHoneywell partner and investor
    Reliable RoboticsAutomationAutomated flight-control systems~$160M raised; FAA-certified automation work
    Iris AutomationDetect-and-avoid"Casia" machine-vision collision avoidance for BVLOS dronesVision-first; lighter recent funding
    SkyryseAutomation"SkyOS" flight automation + "Skylar" AI assistantAutomation-first; AI assistant newer
    Overwatch AIFrontline ops copilot"Wingman" AI assistant for pilots, cabin crew, and ops teams$1.5M pre-seed (2026); supports 30,000+ flights per month
    Neuron InnovationsSurveillance data infrastructureDistributed surveillance as a service for AAM and dronesNATO DIANA cohort; won Revolution.Aero Pitch
    WingbitsSurveillance data infrastructureDePIN network of cryptographically signed ADS-B receivers$5.6M raised in January 2025
    Vigilant AerospaceDetect and avoidFlightHorizon detect and avoid system for BVLOS and AAMUS federal deployments; FAA Part 108 alignment
    ElyoConversational travel searchFlight discovery and comparison across flexible dates and destinationsBuilt by one founder; AI chat replacing OTA date pickers

    Flight Operations, Routing, and Air Traffic Management

    Air Space Intelligence (ASI) is the standout here. Its Flyways platform applies machine learning to flight planning and route optimisation, weighing traffic, weather, and airspace constraints to recommend more efficient routings — Alaska Airlines has been a customer since 2021. In 2026 ASI won the FAA's "SMART" predictive air-traffic-management programme, a contract reported at around $875 million over twelve years, reportedly beating Palantir and Thales (Aviation Week, FlyingMag). That is one of the clearest signals yet that ML is moving into the operational core of the national airspace system, not just back-office analytics.

    A newer entrant in the same broad space is Overwatch AI, which builds a frontline decision support assistant, sometimes referred to as "Wingman," for pilots, cabin crew, engineers, and operations managers. Where ASI optimises the plan, Overwatch focuses on the humans executing it, pulling operational data together so teams spend less time chasing answers between systems. The company came out of Techstars in 2025 and closed a $1.5 million pre-seed round in May 2026, with the platform reportedly supporting more than 30,000 flights per month (Techstars, Tech.eu). Founder and CEO Leo Kotil, a former airline and business aviation pilot, explained on the Travel Tech Podcast why the bottleneck in the cockpit is retrieval under pressure, not a lack of digitisation.

    Cockpit Copilots

    Beacon AI builds "Murdock," an AI copilot designed to work alongside human pilots to reduce error and workload, paired with a flight-data platform it calls "Lighthouse." The company raised a $15 million Series A led by Costanoa Ventures, with backers reported to include JetBlue Ventures and Sam Altman, and in 2026 announced a $49.5 million contract with US Special Operations Command (Aviation Week, PR Newswire). The cockpit is one of the most safety-critical and regulated environments in transport, which makes assurance — not just capability — the gating factor for products like this.

    Autonomy, Detect-and-Avoid, and Perception

    This is the densest segment, and the one where "AI model builder" genuinely applies.

    • Shield AI develops "Hivemind," an autonomous AI pilot for defence platforms, and reached a valuation of roughly $12.7 billion after a 2026 raise (TechCrunch). It is the highest-profile pure-play autonomy company on this list.
    • Merlin Labs builds AI autopilot software intended to retrofit existing aircraft, and went public for around $200 million in 2026 (Washington Technology).
    • Daedalean develops machine-learning and computer-vision avionics for autonomous flight, and counts Honeywell as both a partner and an investor (Venturelab) — genuine perception-model engineering.
    • Reliable Robotics builds automated flight-control systems and has raised around $160 million; its work skews more toward automation and certification than novel ML models (GovConWire).
    • Iris Automation offers "Casia," a camera-and-machine-vision detect-and-avoid system for beyond-visual-line-of-sight drone operations.
    • Skyryse builds "SkyOS," a universal flight-automation system, with an "Skylar" AI assistant layered on top; its strength to date is automation, with the AI assistant a newer addition.
    • Vigilant Aerospace builds FlightHorizon, a detect and avoid platform that fuses transponders, radar, and telemetry into a single airspace picture for BVLOS and advanced air mobility operators. Co-founder and CEO Kraettli Epperson walked through the FAA Part 108 landscape and why detect and avoid is the gating technology on the Travel Tech Podcast.

    Surveillance Data and Airspace Infrastructure

    Every AI system above depends on knowing where aircraft actually are. Traditional surveillance feeds are concentrated in a small number of commercial providers and a volunteer ADS-B community that receives no compensation. Two companies are rebuilding that layer with different models.

    Neuron Innovations, based in London, is developing a distributed surveillance as a service platform for advanced air mobility and drone operations, combining ground based receivers and edge compute into a resilient network. It won the Revolution.Aero Pitch competition for its distributed AAM system and has been analysed as part of the NATO DIANA cohort for resilient C4ISR and European defence autonomy (Unmanned Airspace, Defence Finance Monitor). Founder James Dunthorne, who holds a PhD in aeronautical engineering and also runs 4DSKY, made the case on the Travel Tech Podcast that flying drones beyond visual line of sight is an infrastructure problem before it is an AI problem.

    Wingbits takes a DePIN approach: a global community runs cryptographically signed ADS-B receivers and earns rewards for contributing verified flight data, with the network selling that data into aviation operations. Wingbits raised $5.6 million in January 2025 to scale the network (CoinDesk). Neither company is training a foundation model, but both matter to aviation AI because model quality is bounded by the quality and coverage of the underlying surveillance data.

    Conversational Travel Search and Distribution

    Aviation AI is not only about aircraft and airspace; it also reaches the consumer layer of flight search and distribution. Elyo is built by a single founder as a conversational travel assistant that replaces traditional OTA date pickers with natural language search. Instead of forcing travellers into fixed origin, destination, and date fields, Elyo decomposes intent: flexible weekends, cheapest meeting points for groups, or destination by price discovery. It then queries live flight data to render visual comparisons. Founder Christopher Olivares discussed the architecture, OTA economics, and the GDS access barriers facing early-stage travel AI on the Travel Tech Podcast.

    Predictive Maintenance and MRO

    Here the honest answer is that incumbents lead, not startups. The most mature predictive-maintenance AI sits inside Airbus Skywise, Rolls-Royce's IntelligentEngine, and airline programmes like Delta's APEX. Among smaller players, Finland's QOCO works on MRO data and AI, though independent corroboration is lighter. If predictive maintenance is genuinely the largest AI-in-aviation application by spend, it is also the one where the buyer is most often a manufacturer or major carrier rather than a venture-backed newcomer.

    Aviation LLMs and Foundation Models — Read This Before You Believe a Pitch

    The query "top startups building aviation foundation models" deserves a blunt answer: this category is largely research-led, not startup-led. The most credible aviation-specific large language model, AviationGPT, was built by MITRE — a federally funded research centre, not a startup — by fine-tuning open models such as LLaMA-2 and Mistral on aviation corpora (arXiv). MosaicATM has run FAA-funded feasibility work on aviation foundation models for voice and ATC analytics (MosaicATM). If a vendor claims to have a proprietary "aviation foundation model" and cannot tie it to a programme like these, be sceptical — and ask to see the evaluation evidence.

    Founders on the Record: Every Company from the Travel Tech Podcast

    The list above is built from public sources. This one is built from conversations. Each company below has had its founder, CEO or a senior leader on the Travel Tech Podcast, and the summary of each comes from that episode's show notes. Not all of them are aviation AI companies in the narrow sense. Some sell data, some sell distribution, one makes building materials. We have kept them all, because the same questions about data, trust and deployment run through every conversation.

    CompanyWhat they doGuestEpisode
    Overwatch AIAI assistant for pilots, crew and ops teamsLeo Kotil, Founder & CEOEp 3
    ForthcodeIn-flight retail, catering and an AI agent for ground staffAjith Balakumar, CEO & FounderEp 31
    CiriumAviation analytics and dataKevin Hightower, Chief Innovation OfficerEp 37
    Neuron InnovationsDistributed surveillance for drones and AAMJames Dunthorne, FounderEp 2
    ANRA TechnologiesDrone traffic management, UTM and U-SpaceAmit Ganjoo, Founder & CEOEp 6
    Vigilant AerospaceDetect and avoid for BVLOS and AAMKraettli L. Epperson, Co-Founder & CEOEp 12
    AvtrainDrone training and certificationJulie Garland, CEOEp 25
    Blue Spirit AeroHydrogen electric trainer aircraftEric BidingerEp 38
    RouteMeVideo based indoor wayfinding for airportsDustin Gimbel, Co-founderEp 4
    Smarter AirportsAIRHART airport operations platformMartin Bowman, Chief Strategy OfficerEp 7
    Curie TechnologiesUptime and coordination for airport security equipmentAnne Marie Pellerin, CEO & Co-FounderEp 11
    threedotTravel consultancy on standards and distributionEric Leopold, CEOEp 9
    TWAIAirline retailing and distribution technologyJim Hetzel, Retailing StrategyEp 16
    LeapShiftAirline retailing and commercial strategy consultancyAnn Cederhall, Co-founderEp 22
    MovmoIdentity and express checkout for airlines and hotelsKameron Bertine, Co-founder & CEOEp 29
    CaraveloAirline flight subscriptions and last minute passesIñaki Uriz, Co-founder & CEOEp 32
    TravelDaxFlight pricing intelligence from metasearchValentino Portnoy, CEO & FounderEp 35
    ElyoConversational flight searchChristopher Olivares, FounderEp 17
    Custom Travel SolutionsBooking infrastructure for AI and human agentsMike Putman, CEO & FounderEp 13
    BoopSocial travel itineraries from your camera rollNancy Li Smith, CEO & FounderEp 23
    AmgineAI middleware for corporate travel bookingCaitlin Gomez, Head of GrowthEp 30
    Zoftify and ToursetaTravel software agency and a tour operator platformAlex Ragin, FounderEp 20
    SmartvelTravel content and traveller documentationÍñigo Valenzuela, FounderEp 24
    Swift Passport & Visa ServicesPassport and visa processingRob Lee, Co-Founder & CEOEp 33
    HotelPORTHotel content governance and distributionFred Bean, Founder & CEOEp 15
    Omnitravel and room-matching.comHotel data normalisation and room mappingOlivier Boinet, FounderEp 18
    Visiting MediaHospitality sales platform (TrueTour)Jascha Kaykas-Wolff, CEOEp 39
    Polygraf AIBehavioural security and AI risk managementYagub Rahimov, CEO & FounderEp 21
    LucaBusiness intelligence and capital for small businessesEric Bidinger, FounderEp 38
    SutherlandBusiness and technology services for travelSarah Owen, VP Europe Travel & TransportationEp 25
    PlantdBuilding panels made from fast growing grassJosh Dorfman, Co-founder & CEOEp 19

    Airline Operations and the Front Line

    Overwatch AI. Leo Kotil was an airline and business aviation pilot before he became a founder. His argument is that aviation has already digitised its manuals, Electronic Flight Bag apps and NOTAMs, and that the real problem is finding the right line among thousands of pages while the aircraft is moving. Overwatch works on search latency as a safety issue and on hybrid inference that still works offline. Listen to episode 3.

    Forthcode. CEO and founder Ajith Balakumar started with a point of sale system for a 2,500 outlet coffee chain that also sold on board aircraft, which forced the company to build for zero connectivity years before edge computing was fashionable. Forthcode now sells in-flight retail, catering management, a patented loyalty concept called Lounge in the Sky, and Davix, an AI agent for airport ground staff that cites the SOP page and paragraph behind every answer. It hosts its models locally so airline data never leaves the carrier. Listen to episode 31.

    Cirium. Not a startup, but Chief Innovation Officer Kevin Hightower runs Propulsion Labs, its innovation group, like one. He came from oceanic air traffic control engineering at Lockheed Martin and four and a half years on Google's drone delivery programme. He describes Cirium's aviation knowledge graph, which links aircraft orders, lessors, schedules and tail swaps, and argues that AI will speed up flow management and sequencing long before it touches safety critical separation. Listen to episode 37.

    Airspace, Drones and New Aircraft

    Neuron Innovations. Founder James Dunthorne has a PhD in aeronautical engineering and a background in collision avoidance research and sensor fusion. He explains why low altitude transponder coverage is fragmented, and why edge networks of ground receivers have to exist before AI applications for drones can be trusted. He also runs 4DSKY. Listen to episode 2.

    ANRA Technologies. Founder and CEO Amit Ganjoo is a long standing leader in drone traffic management, UTM and U-Space. He came to aviation from telecoms, and his lesson from that industry's move away from fragmented standards is that complex systems fail at their interfaces. He covers interoperability, certification, and how to move airspace systems from experiment to operation. Listen to episode 6.

    Vigilant Aerospace. Co-founder and CEO Kraettli L. Epperson builds the detect and avoid layer that lets autonomous aircraft share airspace. He explains what FAA Part 108 enables, why aircraft that do not broadcast their position are the hard case, and what a fully integrated airspace could look like by 2035. Listen to episode 12.

    Avtrain. CEO Julie Garland is also president of the Joint European Drone Associations. She began as an aircraft maintenance engineer with Aer Lingus, became a commercial captain and type rating instructor, then qualified as a barrister in aviation law before moving into drone regulation. Her view is that Europe regulates first while the US and China operate first, and that missing planning, noise and airspace rules are sending European drone operators abroad. Listen to episode 25.

    Blue Spirit Aero. This Toulouse company is developing a hydrogen electric trainer aircraft for professional pilot schools. Eric Bidinger, an engineer who spent about 17 years in private equity before turning entrepreneur, explains why certification covers the whole design process and not just the finished airframe, and why long gaps between clean sheet programmes have let design knowledge retire with senior engineers. Listen to episode 38.

    Airports

    RouteMe. Co-founder Dustin Gimbel turned a pilot at Miami International Airport into a multi year contract and airline app integrations. RouteMe uses web based video routing in place of beacons and dedicated apps, which matters in multi level concrete buildings where GPS does not work. One operational result he reports is lower demand for reduced mobility assistance. Listen to episode 4.

    Smarter Airports. Chief Strategy Officer Martin Bowman has more than 25 years in aviation and technology. The episode centres on Heathrow adopting the AIRHART platform as its operational backbone, and on why airports are moving away from the traditional Airport Operations Database towards configurable platforms they control themselves. Listen to episode 7.

    Curie Technologies. CEO and co-founder Anne Marie Pellerin is a former TSA leader who was Director of Checkpoint of the Future and spent six years as the agency's representative in Europe. Curie works on uptime for airport security equipment. Her point is that one failed machine becomes queues, baggage disruption and delays because coordination is poor, not because detection is. Listen to episode 11.

    Airline Retailing, Distribution and Pricing

    threedot. CEO Eric Leopold spent 15 years at IATA developing industry standards, including the move from magnetic stripe to barcode boarding passes. He argues that the next wave of travel innovation will be limited by data consistency, identity and industry alignment more than by technology, and that AI agents will need frameworks for identity verification and dispute resolution. Listen to episode 9.

    TWAI. Jim Hetzel leads retailing strategy at TWAI after a career that runs from corporate travel agencies to enterprise distribution platforms. He makes the case that NDC is a messaging standard and not a retailing platform, and that AI can sit across both legacy GDS infrastructure and NDC content as a normaliser. He also names the unsolved problem: nothing yet authenticates machine to machine agent transactions the way IATA certified airlines and agents. Listen to episode 16.

    LeapShift. Co-founder Ann Cederhall wrote the State of Airline Retailing 2026 report and has held senior roles at Lufthansa, Scandinavian Airlines, Amadeus and ATPCo. She delivered Lufthansa's original NDC Direct Connect in 2015. She draws a clear line between what agentic tools can speed up and what they cannot fix, such as shopping engines that have gone 30 years without modernisation. Listen to episode 22.

    Movmo. Co-founder and CEO Kameron Bertine founded Movmo in June 2024 and modelled it on Shopify ShopPay, Stripe Link and Klarna. It sits as an overlay on whatever PSS, GDS or NDC stack an airline already runs and leaves the airline as merchant of record. He puts airline cart abandonment at roughly 89%, and is staging the rollout over two to three years before agents book on their own. Listen to episode 29.

    Caravelo. Co-founder and CEO Iñaki Uriz runs the platform behind flight subscription and last minute inventory products for carriers including Wizz Air and JetSmart. He says that for every $100 of subscription revenue, $30 to $50 is money the airline would not otherwise have captured, and that a private subscription deal is the one thing a fare comparison agent can never find. Listen to episode 32.

    TravelDax. CEO and founder Valentino Portnoy is a software engineer who built one of the first metasearch engines in Latin America. TravelDax collects flight pricing data from metasearch and sells competitive intelligence to airlines and OTAs. In its own testing, classic machine learning still beats generative AI for price prediction. Listen to episode 35.

    Travel Search, Booking and Traveller Services

    Elyo. Christopher Olivares is the solo founder. He spent four and a half years at the OECD in Paris and had never written code before building Elyo with generative AI. The product takes no commissions and is built around traveller intent, such as the cheapest place for a group to meet. Listen to episode 17.

    Custom Travel Solutions. CEO and founder Mike Putman launched one of the earliest online travel agencies in the mid 1990s and now builds infrastructure for AI driven booking. He also built reconfirm.ai, which places AI calls to hotels before arrival to confirm reservations, after finding that roughly two percent of bookings quietly disappear inside distribution systems. Listen to episode 13 and episode 26.

    Boop. CEO and founder Nancy Li Smith previously led the AI platform behind the Meta Ray-Ban glasses and ran perception AI and augmented reality partnerships at Microsoft. Boop turns a camera roll into a bookable itinerary by processing photo metadata on the device, and splits affiliate commission 50/50 with the person who created the trip. Listen to episode 23.

    Amgine. Head of Growth Caitlin Gomez describes Amgine as middleware that automates the manual parts of corporate travel booking across email, Slack and WhatsApp without displacing the GDS and TMC infrastructure underneath. Her career covers TMC sales, partnerships at Lyft and CLEAR, and founding Innovation Row at GBTA. BCD made a strategic investment after four months of results. Listen to episode 30.

    Zoftify and Tourseta. Founder Alex Ragin has built travel software since 2015. Zoftify is a development agency of around 35 people, and Tourseta is a booking and operations platform for high volume multi day tour operators. His rule of thumb is that code is about 20% of complex software and the rest is domain knowledge, process and edge cases. Listen to episode 20.

    Smartvel. Founder Íñigo Valenzuela has run Smartvel for over 13 years, serving airlines, airports, cruise lines and OTAs including Iberia, Delta, United, Ryanair and MSC. The company began with destination events content, pivoted during COVID to tracking travel restrictions, and now covers visas and health documentation. He thinks the underrated legal risk in AI travel content is image licensing, not hallucination. Listen to episode 24.

    Swift Passport & Visa Services. Co-founder and CEO Rob Lee started the business in a Chicago apartment in 2008. Swift keeps customer data out of public AI tools and uses AI internally to scrape government and consulate sites into a live requirements database that a person reviews. He explains why general purpose models give wrong visa guidance. Listen to episode 33.

    Hospitality Data

    HotelPORT. Founder and CEO Fred Bean launched HotelPORT in 2019 after three decades in hotel reservations, GDS connectivity and online distribution at Hyatt, Sabre and TravelWeb. An audit across 2,000 hotels found $7 million in uncollected OTA virtual card payments. His position is that governed data comes before any useful AI in hospitality. Listen to episode 15.

    Omnitravel and room-matching.com. Founder Olivier Boinet has 30 years in software, from antivirus heuristics to natural language processing. room-matching.com maps identical hotel rooms that suppliers list under different names, and Omnitravel prepares hotel and tour operator data for AI. He argues that content written for both people and LLM crawlers is now a distribution strategy. Listen to episode 18.

    Visiting Media. CEO Jascha Kaykas-Wolff came from Yahoo, Microsoft, BitTorrent and Mozilla, where he was CMO. Visiting Media's TrueTour platform converts immersive property content into written descriptions and FAQs that answer engines can read. He also describes rebuilding the company's own work around 17 agents tied to departments. Listen to episode 39.

    Security, Enterprise AI and the Wider Field

    Polygraf AI. CEO and founder Yagub Rahimov works on behavioural security and contextual privacy for defence, financial services and technology clients. He treats AI agents as a new kind of user with employee level access and no training against deception, and reports a 72 percent reduction in data loss prevention violations for one client through real time intervention. Listen to episode 21.

    Luca. Eric Bidinger's second company combines business intelligence with access to capital for small and medium businesses. He describes a four layer enterprise AI stack in which ordered data, memory and human taste matter before model choice does. Listen to episode 38.

    Sutherland. Sarah Owen is VP for Europe Travel and Transportation and chairs the Aviation Club of the UK Forum. She has held senior roles at Google, Salesforce and Cyient, and led commercial strategy at Assaia, a computer vision company focused on aircraft turnaround. She talks about the data silos that stop airports, handlers and airlines from sharing operational intelligence. Listen to episode 25.

    Plantd. Not a travel company at all. Co-founder and CEO Josh Dorfman makes structural building panels from perennial grass in North Carolina. He is here for what he says about sustainability that customers will actually buy, and for the Jefferies estimate that weight loss drugs could save US airlines around $580 million a year in fuel. Listen to episode 19.

    From a List to a Working System

    Every company above is building capability. Two questions follow each of them into deployment: does the system work on real operational data, and can you show that it fails safely?

    That is the work Airside Labs does. We build AI prototypes for aviation and travel teams on their own data, and we test them before anyone decides to scale. Two pieces of that are open to everyone:

    1. Agent tools. Our aviation tools for AI agents connect Claude, ChatGPT or your own agent to a catalogue of 6,527 aviation AI use cases and to lookups that resolve airports, airlines, aircraft and flights correctly for a given date.
    2. Domain competence. The open-source Pre-Flight benchmark gives a fast, like-for-like baseline of how well a model understands aviation, which is useful when you are selecting or comparing models.

    If you are building or buying aviation AI and want a working prototype instead of a slide deck, start a conversation.

    Works cited

    1. Air Space Intelligence raises $34M to run flight ops — Aviation Week
    2. ASI wins FAA SMART air-traffic-management contract — FlyingMag
    3. Beacon AI raises $15M for AI co-pilot — Aviation Week
    4. Beacon AI Series A — PR Newswire
    5. Shield AI $12.7B valuation — TechCrunch
    6. Merlin Labs public offering — Washington Technology
    7. Daedalean — Venturelab
    8. Reliable Robotics autonomy system — GovConWire
    9. AviationGPT — arXiv
    10. MosaicATM aviation foundation models — MosaicATM
    11. AI in Aviation market size — Fortune Business Insights

    Frequently asked questions

    What are the top startups building aviation AI in 2026?

    Well-corroborated aviation AI builders in 2026 include Air Space Intelligence (ML flight operations and air traffic management; won the FAA's SMART programme), Beacon AI (cockpit copilot, with a US SOCOM contract), Shield AI (Hivemind autonomous AI pilot, valued around $12.7B), Merlin Labs (AI autopilot software, now public), and Daedalean (machine-vision autonomous-flight avionics, partnered with Honeywell). Reliable Robotics, Iris Automation, and Skyryse are notable in autonomy and detect-and-avoid, though their work skews more toward automation than novel model-building.

    Which travel and aviation tech founders have been on the Travel Tech Podcast?

    Guests include the founders or leaders of Overwatch AI, Neuron Innovations, ANRA Technologies, Vigilant Aerospace, Avtrain, Blue Spirit Aero, RouteMe, Smarter Airports, Curie Technologies, threedot, TWAI, LeapShift, Movmo, Caravelo, TravelDax, Elyo, Custom Travel Solutions, Boop, Amgine, Zoftify and Tourseta, Smartvel, Swift Passport & Visa Services, HotelPORT, Omnitravel, Visiting Media, Polygraf AI, Luca, Forthcode, Cirium, Sutherland and Plantd. Each is profiled in this guide with a link to the episode.

    Are there startups building aviation-specific foundation models or LLMs?

    This category is largely research-led rather than startup-led. The most credible aviation-specific large language model, AviationGPT, was built by MITRE — a federally funded research centre — by fine-tuning open models such as LLaMA-2 and Mistral on aviation data. MosaicATM has run FAA-funded work on aviation foundation models for ATC analytics. If a vendor claims a proprietary aviation foundation model without ties to programmes like these, treat the claim with caution and ask for evaluation evidence.

    How do you evaluate whether an aviation AI startup's models are safe?

    Independently, in two layers. First, baseline the model's aviation domain competence — the open-source Pre-Flight benchmark scores a model against a human-expert reference. Second, test adversarial robustness and compliance with use-case-specific red-teaming mapped to the EU AI Act, OWASP LLM Top 10, NIST AI RMF, and MITRE ATLAS. Airside Labs builds aviation AI prototypes and tests them this way before they are scaled.

    Airside Labs Team

    Airside Labs Team

    Research & Development

    The Airside Labs team comprises aviation experts, AI researchers, and safety-critical systems engineers dedicated to advancing AI evaluation methodologies. Our collective expertise spans air traffic management, ground operations, commercial aviation, and AI security.

    Newsletter

    Every episode, in your inbox

    One email per Travel Tech Podcast episode: the conversation, the key takeaways and the use cases worth building in aviation and travel AI.

    Agent Tools

    Give your agent aviation tools

    General assistants guess at aviation codes and get them wrong. Agent Tools give Claude, ChatGPT or your own agent checked answers instead. Free plan, no card.

    Connect your agent