ANA puts an AI turbulence forecast into its flight operations
ANA says it is the first airline to use BlueWX's deep learning turbulence prediction in operations, after trials with 2,500 of its pilots since 2021. The model came out of ANA's own research with Keio University. Remote support robots at regional airports are staffed by people and stay in the wider register.
The operating picture
A forecast built inside the airline
All Nippon Airways Boeing 777. Photograph: Masahiro TAKAGI / CC BY 2.0 (cropped to fit). Illustrates the airline, not an AI deployment.
Clear-air turbulence does not show on onboard weather radar, so crews rely on forecasts and on reports from aircraft ahead. ANA and Keio University began developing a deep learning turbulence model in 2019, started trial use in 2021 and set up BlueWX as a company in July 2023. In August 2025 ANA said the forecast had been formally added to its weather data systems.
ANA's second record is a service experiment. Since August 2026, newme telepresence robots at regional airports have connected passengers to staff working from Kansai International (KIX). The robots are operated remotely by people and the release names no AI in the passenger service, so that record stays in the wider register.
Selected implementations
The work behind the evidence
01 / Deployed
A turbulence model trained on ten years of data
The operational problem
Dispatchers and pilots plan routes and altitudes around expected turbulence, and cabin crew need warning to secure the cabin. A forecast that misses rough air, or flags it too often, soon loses the trust of the crews using it.
What the evidence reports
ANA said on 7 August 2025 that it had become the first airline to put BlueWX's AI turbulence prediction into operation. The deep learning model is trained on about a decade of turbulence data, and ANA reported 86% accuracy after operational trials involving 2,500 of its pilots. Hiroyuki Kometani, executive vice president of ANA's operation division, said the capability was developed with BlueWX "utilizing the latest AI and deep learning techniques".
Operational significance
The trial ran for four years with the pilots who would use the output, which is how a forecast earns crew confidence. Because BlueWX was founded by ANA and Keio University, the airline is both a developer and the first operational customer.
Evidence still needed
ANA has not said how the 86% figure is measured, what lead time or area the forecast covers, or how forecasts reach dispatchers and pilots. No figures on turbulence encounters or injuries since adoption have been published.
Other records are shown separately, including non-AI technology and programmes at different stages of deployment. A register entry does not establish a measured AI benefit.
avatarin newme telepresence robots for ANA remote passenger support · Deployed · limited scope
ANA and avatarin began rolling out remote passenger support at regional airports on 21 August 2026, using newme telepresence robots operated from Kansai International Airport.
In earlier testing at Haneda, 98.4% of enquiries were resolved remotely.
Tottori Airport became the third airport to use one, on 1 October 2026.
ANA's turbulence forecast is one of the few airline AI deployments aimed directly at flight safety and cabin comfort, and it was built and trialled at scale inside the airline before it went into service.
Its value will be settled by operational evidence: turbulence injuries, interrupted cabin service and altitude changes before and after adoption, alongside a published definition of the accuracy figure. BlueWX's plan to offer the service to other airlines may produce that comparison.
Live data from the Aerostack register. Status reflects published evidence only. See Our approach for how records are graded.
Aerostack register · 2 tracked deployments
Aerostack tracks AI deployments across airports and airlines, with the published evidence behind each record. Photography is real aviation imagery. Record thumbnails may show the named airport or airline; other photographs do not.
_at_Tokyo_Haneda_Airport.jpg)