AFI KLM E&M
AFI KLM E&M is the maintenance, repair and overhaul arm of Air France-KLM. Its PROGNOS suite applies machine learning to flight, taxi and gate data to predict component failures before they ground an aircraft.
Formed in 1994, the Paris-based organisation has more than 10,000 staff. It runs PROGNOS with its auxiliary power unit specialist EPCOR and has used it on Air France-KLM fleets since 2016. Other tools include TRAK, which finds resolution procedures across manuals and bulletins, Voice to Admin for spoken shop-floor reporting, and LiMiTS for fitting line maintenance into ground times.
Akasa Air signed in 2023 to use PROGNOS on the auxiliary power units of its 737 MAX 8 fleet, the first introduction in India. The company says the suite monitors thousands of aircraft for more than 75 operators and predicts failures 10 to 50 flights ahead; those figures are its own. A 2025 framework with Delta, Virgin Atlantic and IndiGo covers maintenance cooperation, not PROGNOS.
Products & Solutions
5 productsTRAK (Transparent Reliable Accessible Knowledge)
TRAK stands for Transparent Reliable Accessible Knowledge. It is a generative AI-powered knowledge management system developed by AFI KLM E&M (KLM Engineering & Maintenance) specifically to reduce search and decision time for structure engineers and technicians in Aircraft on Ground (AOG) situations.
In AOG situations, KLM E&M engineers must locate correct resolution procedures across thousands of internal manuals, OEM bulletins, and historical case records, often under time pressure and without access to the most experienced engineers. TRAK consolidates this documentation into a single GenAI-searchable system. The secondary concern it addresses is institutional knowledge retention: as experienced engineers retire or leave, their working knowledge is preserved within the system rather than lost.
- Dedicated GenAI tool for resolving AOG (Aircraft on Ground) situations by consolidating tens of thousands of relevant maintenance documents into a single searchable tool
- Uses GenAI to automatically generate short summaries and standardized titles for individual documents, enabling engineers to skim results faster than reading full-length manuals
- Addresses knowledge-gap risk caused by retirement and staffing shortages of experienced senior engineers by centralizing institutional knowledge
- New AOG case knowledge can be added to the system using speech-to-text technology
1 more feature
- Developed and maintained within the KLM E&M Engineering division; at Deploy & Scale stage in the internal GenAI use case funnel as of June 2024
Technical specifications (3)
- GenAI application based on large language model (LLM) summarization of internal documentation corpus
- AFI KLM E&M industrialized GenAI use cases including TRAK on a dedicated platform built in partnership with Google, enabling benchmarking of different models (e.g., ChatGPT, Google Gemini)
- No specific AI model version, vector database, or retrieval architecture explicitly named in sourced documents
KLM Engineering & Maintenance (2024)
Customers and sites the vendor lists for this product. Aerostack has not verified these unless a deployment record is linked above.
Product documentationCharlie
Charlie is a generative AI-powered tool developed by Air France Industries / AFI KLM E&M for use by aircraft maintenance technicians.
Maintenance technicians searching for the correct part number and associated repair task must manually cross-reference OEM documentation and internal maintenance manuals. This is a time-intensive step in each repair or replacement event. Charlie uses GenAI to retrieve the relevant part number, tasks, and documentation on demand. Aviation Week, citing AFI KLM E&M internally, reports a saving of over one hour per event on average.
- GenAI-powered search tool that lets maintenance technicians instantly find correct parts, tasks, and essential documentation within airline and OEM maintenance records
- Enables technicians to identify correct part numbers and order spare parts more quickly when a component is broken or needs replacement
- Reduces the time required to repair or replace aircraft parts by more than one hour on average, per AFI KLM E&M
- Deployed for use by maintenance teams within the AFI KLM E&M Component Services (CS) repair shop environment
Technical specifications (2)
- GenAI-based document search application; no specific AI model type, architecture, or underlying platform is explicitly identified in sourced documents
- Searches across airline internal documentation and OEM (Original Equipment Manufacturer) maintenance documentation
Air France maintenance teams (2025) · KLM maintenance teams (2024)
Customers and sites the vendor lists for this product. Aerostack has not verified these unless a deployment record is linked above.
Product documentationVoice to Admin
Voice to Admin is a natural language processing and speech-to-text tool developed by AFI KLM E&M for its Component Services (CS) repair shops, specifically prototyped for the E&M Hydraulics Shop.
Technicians in MRO repair shops are required to stop hands-on work in order to manually enter maintenance findings into ERP systems. Where the technician works in a language other than the required system language, this introduces additional friction and inconsistency in the data recorded. Voice to Admin allows technicians to speak their findings in Dutch, English, or French; the system transcribes, translates to English, validates the input format, and writes it to the ERP without manual intervention.
- NLP speech-to-text tool that allows MRO repair shop technicians to verbally record maintenance data, eliminating handwritten notes and manual ERP data entry
- Translates spoken input in the technician's native language into English for entry into the company's ERP system
- Includes a proofreading step that flags ambiguous or non-standard descriptions (e.g., prompts technician to clarify 'broken' as 'dented' or 'scratched') and ensures standard format
- Reduces data-quality errors caused by illegible handwriting, shift-handover miscommunication, and inconsistent entry formats
3 more features
- Resulted in higher shop productivity and better data quality per AFI KLM E&M
- Piloted in the CS (Component Services) Repair Shop and prototyped for the E&M Hydraulics Shop
- Recording supported in Dutch, English, and French
Technical specifications (4)
- NLP-based speech-to-text pipeline with output standardization and proofreading logic
- Supports recording input in Dutch, English, and French; translates to English for ERP report generation
- Output is ingested into the company's ERP system; no specific ERP product name identified in sourced documents
- No specific underlying speech recognition engine or LLM model named in sourced documents
KLM E&M Hydraulics Shop (2024)
Customers and sites the vendor lists for this product. Aerostack has not verified these unless a deployment record is linked above.
Product documentationPROGNOS
PROGNOS is AFI KLM E&M's flagship predictive maintenance suite, the company's commercially headlined AI/big-data product line, marketed to over 80 external airline customers worldwide. The suite is organised into four modules covering Aircraft, Engine, APU, and Inventory.
Unscheduled in-service failures of aircraft, engine, APU and rotable components drive AOG events, flight delays, and high inventory carrying costs. PROGNOS applies ML models to operational and sensor data to forecast component failures 10–50 flights in advance, allowing planners to schedule replacement work and stock the right parts before failures cause disruption.
Air France · KLM
Customers and sites the vendor lists for this product. Aerostack has not verified these unless a deployment record is linked above.
LiMiTS
LiMiTS is an AI-powered line maintenance scheduling tool developed within KLM Engineering & Maintenance. It went live in June 2024.
Line maintenance work must be fitted into short aircraft ground times at the gate without disrupting departures. Manual scheduling makes it hard for planners to see whether a proposed plan is actually feasible against ground time and crew availability. LiMiTS uses AI to optimise gate maintenance scheduling and surface feasibility constraints up front.
KLM Engineering & Maintenance (2024)
Customers and sites the vendor lists for this product. Aerostack has not verified these unless a deployment record is linked above.
TRAK (GenAI AOG Resolution)
Profile data from the Aerostack companies register; deployment records from the verified register. Evidence grades: T1 first-party confirmation, T2 credible reporting. See Our approach for how records are graded.