Over the past few years, the conversation about artificial intelligence in aviation has shifted from whether the technology will matter to where, in practice, it is already being deployed. Activity is now visible across operations, maintenance, passenger services, air traffic management and regulation, although the pace of adoption varies considerably between them, shaped above all by how safety-critical each activity is and by how much regulatory approval any change requires.
ALG’s Digital & AI expert team has tracked more than 300 Data and AI use cases across the aviation value chain, spanning airlines, airports, air navigation service providers, authorities, ground handling and MRO. This radar sets out seven of them, selected to describe where the sector stood at the end of June 2026. Every one is a real initiative, either already in operation or publicly announced, with the organizations responsible for it named.
The seven cases are not intended to represent every application of Data and AI in aviation, nor were they chosen simply because they are the most advanced. Taken together, they show which parts of the business are already seeing real deployment, how far that deployment has progressed, and how much confidence the published results warrant, which is the distinction that tends to matter most when deciding where to invest next.
What this snapshot tell us
1- Progress is fastest at the two ends of the business
The most advanced cases are concentrated in ground operations, where the cost of a delay is felt immediately, and in revenue management, where a better pricing decision can quickly be seen in the numbers. Adoption is slower where the benefit is real but harder to link to a single cause, as is often the case in safety and sustainability. The difficulty there is usually in building the business case, rather than in the technology itself.
2-Past proof of concept, short of scale
Almost all of the cases sit in the middle of the maturity range, which means they are working systems rather than trials. Even so, only two of the seven have been extended across an entire fleet or network, while the rest operate well at a limited number of sites, leaving the move from a few successful locations to consistent network-wide operation as the harder part still ahead.
03- The main constraint is data, not algorithms
Several of the most advanced cases are, in substance, data integration programmes. The organizations deploying AI at pace tend to be those that addressed their data foundations first, connecting previously fragmented sources and making the information reliable enough to use.
04- Independently confirmed results remain limited
Only two of these cases carry results confirmed by an operator or a regulator. We label every figure by its source, and we recommend the same discipline when assessing supplier proposals.
The Q2 2026 radar
The radar is organized into six segments, one for each part of the aviation business: operations, safety, enterprise effectiveness, sustainability, asset management and commercial. Each case appears as a marker within the segment it serves, positioned by maturity, so that early-stage work sits closer to the center and deployments already in full production sit towards the outer edge. Reading a segment from the inside out therefore gives a sense of how far that part of the business has progressed.
The color of each marker indicates the stakeholder the case principally serves, whether an airline, an airport or an air navigation service provider, and the numbers correspond to the seven cases described in the following section, which examines each one in turn.
Seven deployments in 2026, in detail
The seven cases follow. Each one sets out the problem it addresses, what has actually been built, where the deployment stands today, and our assessment of what it means, with the source of every figure identified.
Computer vision for aircraft turnaround monitoring
The problem: A turnaround is a tightly choreographed sequence of chocks, baggage, refuelling and catering, spread across several teams. When progress is only partly visible in real time, coordination slips and stand capacity is used less efficiently than it could be.
What was built: Video analytics reads live camera feeds around stands and boarding bridges, detecting and time-stamping events such as aircraft positioning, ground power connection, chock placement, refuelling and catering. Operations teams gain a factual, real-time view of every turnaround in progress.
Where it stands: Operating at three Spanish airports following an initial deployment at Madrid-Barajas. Deployment at Madrid-Barajas, Barcelona-El Prat and Palma de Mallorca is confirmed by the airport operator. Quantified gains in stand utilisation or punctuality have not been published, so the benefit case has to be built against an airport's own baseline.
ALG's view: Of all the cases reviewed, this one has the strongest operational record, largely because it adds to infrastructure an airport already has instead of replacing it. Since the analytics run on existing cameras and stands, both the integration cost and the implementation risk stay low, which is why the solution has reached three major airports while more ambitious ideas remain in trials. What will decide any further deployment is practical is, above all, the camera coverage available and the geometry of each stand, since the underlying computer vision is already mature.
Predictive decision support for airspace flow management
The problem: Traffic is above pre-pandemic levels and unevenly distributed, concentrating pressure on particular corridors, airports and control sectors. Balancing that traffic requires processing real-time data at a volume manual coordination was never designed to handle.
What was built: Software forecasts airspace conditions by combining aircraft, airport, weather and network data, then supports controllers in resolving conflicts earlier and rebalancing flows. Delay, cost and workload all fall, while the controller retains the decision.
Where it stands: Competitive evaluation of providers completed; award and deployment scope not yet public. The FAA has run an evaluation of technology providers, with the outcome expected shortly. Because no deployment scope has been published, we place this case in the Trending ring rather than further out.
ALG's view: What matters most in this case is how it is set up, because the system proposes and the controller decides, and that is currently the only arrangement likely to gain approval in controlled airspace. As a result, the real constraint on wider use is assurance and explainability, meaning the ability to show why a recommendation was made and to establish who is accountable for acting on it, and not the accuracy of the prediction itself. Until the FAA announces the award and its scope, the capability claims in this segment cannot be assessed independently, which is why the case sits in the Trending ring.
Predictive analytics for de-icing and winter runway operations
The problem: Winter runway management depends on reading weather forecasts, runway sensors, and traffic schedules at the same time. Without analytical support, the decision becomes reactive: chemicals are over-applied, capacity is lost, and delay and environmental cost follow.
What was built: A model predicts runway surface conditions several hours ahead and simulates them with and without chemical treatment, allowing supervisors to choose the timing and dosage of de-icing deliberately rather than defensively.
Where it stands: Developed with the national airport operator and in use in Norwegian winter operations. Avinor has described the approach and its objectives publicly. Reductions in chemical use and retained capacity have not been quantified in public material, so the benefit case remains directional rather than measured.
ALG's view: The gain here comes from treating the runway less often and more precisely, applying de-icing when the forecast conditions call for it rather than as a precaution. It is also unusually easy to justify internally, because safety, cost and environmental performance all improve from the same model, which keeps the functions that normally own those objectives working towards the same decision. Its applicability is limited to airports with meaningful exposure to winter weather, and that is what defines where it is worth pursuing.
Contrail forecasting integrated into flight planning
The problem: Contrails are the ice-crystal clouds formed when aircraft cross cold, humid air, and they account for a significant share of aviation's warming impact. The effect is short-lived but intense, which makes it a rare near-term lever. The difficulty lies in knowing which flights, at which altitudes, will create high-impact contrails.
What was built: A forecasting model combines satellite imagery, weather and humidity data with historical flight information to map contrail-prone altitude bands and regions. The forecasts feed directly into flight-planning software so that routings can be adjusted before departure.
Where it stands: Forecasts in use within flight planning at one major carrier following a completed trial. A 2025 trial reported a 62% reduction in contrails and a 69% reduction in warming effect on adjusted flights, with no significant fuel increase at trial scale. The share of aviation's warming impact attributed to contrails is an IPCC figure.
ALG's view: This is the only case reviewed in which a climate benefit has been measured rather than modelled, which distinguishes it from most sustainability activity in the sector. The routing adjustments involved are modest. The material uncertainty is the fuel and payload trade-off, immaterial at trial scale but requiring governance at flight level before the approach is extended across a network.
A shared data platform for fleet and maintenance analytics
The problem: Aircraft, maintenance and operational data has historically been fragmented and under-used. Many airlines lack the capability to decode and integrate it at volume, which caps the quality of fleet planning and predictive maintenance and leaves unplanned downtime in the system.
What was built: A cloud platform brings aircraft, maintenance and operational data from multiple stakeholders into one environment, structures it into usable models, and layers analytics, machine learning and applications such as defect analysis and fleet tracking on top.
Where it stands: In service with airline and MRO customers across multiple fleet types. Adoption across a substantial base of airline and MRO customers is documented by the manufacturer. Savings attributable specifically to the platform, as distinct from the maintenance programmes around it, are not separately published.
ALG's view: The analytical capability of a platform like this rests on preparatory work that is easy to underestimate, namely the decoding, structuring and modeling of aircraft data that has to happen before any analytics are possible. The practical implication for a carrier is that the outcome cannot be bought independently of that foundation, and that data preparation usually accounts for the larger share of the total cost, so the return depends far more on the state of the airline's own data than on the platform in isolation. Understood in those terms, the case is as much a question of data readiness as it is a technology choice.
A single order record for airline retailing
The problem: Airline retailing still runs on separate constructs: passenger name records, electronic ticket numbers and ancillary documents. The fragmentation is invisible in an uneventful journey, but it surfaces the moment something is changed or disrupted, slowing rebooking and refunds and limiting what can be automated.
What was built: A single master order identifier represents the whole trip, consolidating flight segments and ancillaries such as seats and bags into one record within an order-based retailing architecture. The result is a cleaner foundation for automation across channels and carriers.
Where it stands:Industry standard in progressive adoption, with early implementations at major carriers. One Order is a data-architecture standard for order-based retailing rather than an AI capability, and we include it on that basis.
ALG's view:This case is included precisely because it involves no artificial intelligence. Commercial models cannot perform better than the reference data available to them, and fragmented order records place a ceiling on what pricing and servicing automation can achieve. It is therefore better understood as a precondition for the commercial cases on this radar than as an alternative to them, which makes its sequencing more consequential than its individual return.
Generative market models for dynamic pricing
The problem: Traditional revenue management works through a limited set of fare classes refreshed a few times a day. Competitor moves, seasonality, local events and booking pace interact continuously, so a structural gap opens between the granularity of the market and the speed at which an airline can respond.
What was built: A generative model trained on aggregated market signals simulates forward market dynamics and publishes optimal fares in real time at product, route and timing level, with explainability provided for revenue management teams.
Where it stands: The technology is reported to be in production with several international carriers, and the supplier reports revenue uplifts in the mid to high single digits. Those figures originate with the supplier, have not been independently audited, and should therefore be read as claims rather than as verified results.
ALG's view: The technical approach is credible, and the reason the case sits where it does is simply that the results cannot yet be verified from outside the company. The more important question for an airline is organizational rather than technical, because adopting this means deciding whether its commercial function, its governance and its data are ready to hand fare setting to a model, and what oversight it keeps once they are. On the evidence available today, the sensible next step is a controlled trial on a defined set of routes, measured against a comparable group, so that the airline establishes the effect for itself before committing further.
What this means
For airlines
The clearest returns on the radar are in revenue management and fleet data. Both depend on the same starting point: aircraft, maintenance and commercial data that has been decoded, connected and made reliable. A pricing or predictive maintenance model can only work with the information beneath it, which is why the data groundwork is often where the first investment needs to go.
Order based retailing also deserves attention in this planning cycle, as the choices made now will shape commercial systems for the coming decade.
For airports and handlers
Turnaround monitoring has progressed from pilot to multi-airport deployment more rapidly than anything else here. It can be added over existing infrastructure without replacing core systems, and it returns value through stand utilisation and punctuality within a single season.
For airports in colder climates, winter operations analytics is the other area where prediction repays investment quickly, since safety, cost and environmental outcomes all improve from a single model.
For ANSPs and authorities
In airspace and other safety-critical operations, artificial intelligence is generally being introduced as decision support, with the controller or supervisor retaining the final decision. That formulation is what renders it approvable, and it should frame how requirements are written.
The more difficult questions concern assurance and explainability rather than predictive accuracy: how a recommendation is supported by evidence, and where accountability sits once action is taken.
How we built this radar
We only include work that is real: initiatives that are either already in operation or have been announced publicly, with the organizations behind them identified. Concepts and roadmaps are left out. Each case is then placed in the segment of the business it affects and in the maturity ring that reflects how far the deployment has progressed in practice.
Every case also carries an evidence label identifying who stands behind the result. Where a figure can be traced to an operator, a regulator or a peer reviewed study, we present it as reported. Where a supplier quotes results for its own product, we make that clear, as the two carry different weight when assessing an investment. Where we could not source a detail, we have left the gap visible.
Every position reflects the state of each deployment as of 30 June 2026, and some will have moved since then. If you think we have placed a case incorrectly, or left out one that belongs here, we would be glad to hear from you.
Conclusion
The seven cases here are a selection. Behind them, ALG maintains a portfolio of more than 300 Data and AI use cases across the aviation value chain, covering airlines, airports, air navigation service providers, civil aviation authorities, ground handling and MRO. The radar shows those that best describe where the sector stood in mid-2026.
For airlines, airports, ANSPs and authorities, the practical question is no longer whether these techniques work but which of them is worth funding next, and what the underlying data has to look like first. Answering it calls for a view of the technology, the operation and the investment case at the same time.
With extensive international experience across the aviation, transport and digital sectors, ALG supports clients in identifying which use cases to pursue, sequencing the data work they depend on, and building the business case that carries them through investment approval. If something on this radar is already on your agenda, we would be glad to talk it through.
About the authors
Miguel San Roman
Industrial Management Engineer with an MSc in Big Data Science, specialized in designing and deploying data science and AI solutions for aviation safety and airport operations. Experienced in advanced analytics and applied artificial intelligence across complex operational environments. Senior Consultant in the Data & AI team at ALG.
Antonio Cabeza
Aeronautical engineer with a MSc in Big Data and Advanced analytics. Specialises in strategical projects in airport and air traffic operations with wide expertise in digitalisation and regulatory related projects. Senior Engagement Manager at ALG.
Núria Alsina Pujol
Aeronautical engineer with specialisation in air navigation and systems, certified in project management and scrum methodology. Principal at ALG and head of digitalisation and advanced analytics in the Transportation practice.
