Faced with a large volume of unstructured contracts, Airbus partnered with CGI to industrialize and improve the reliability of its contractual data usage. The objective was clear: to move from fragmented document management to a structured, consistent database, usable on a global scale.
In a demanding industrial environment, control over contractual information is a strategic lever for management, performance and competitiveness.
The challenge:
Managing a global real estate portfolio based on reliable data
Airbus is present across more than 300 sites and manages more than 2,500 buildings worldwide, representing approximately 11 km² of built surface area.
At this scale, lease contracts govern financial commitments, legal responsibilities and site operating conditions. Consistent management of these agreements is essential to effectively oversee a global real estate portfolio.
However, a significant portion of contractual information remained within unstructured PDF documents. Contract analysis relied primarily on manual reading, which was followed by data entry into spreadsheets. Each contract required an average of three hours of processing.
This approach resulted in:
- Scattered information
- Diverse and varied interpretations of clauses
- Difficulties consolidating data at scale
Over time, this way of working reduced overall visibility and limited opportunities for optimization.
The challenge was to structure these documents to obtain reliable, comparable and searchable data to support management and overall performance.
The solution:
A semantic foundation and an industrialized extraction engine
Airbus and CGI jointly developed an approach based on five principles:
- Unification of semantics
- Automated data extraction
- Consistent user experience
- Integration of information into business tools
- Value creation from contractual data
Airbus Real Estate’s business experts defined the business rules and semantic framework needed to standardize contractual information at the outset.
CGI translated these business rules into an operational AI solution deployed on Google Cloud, structuring the prompts and extraction logic so the system could accurately identify and process the expected contractual data.
The system includes optical character recognition (OCR) and detection of relevant areas, identification of the document type and key dates, extraction and annotation of clauses, generation of a summary, and assignment of a confidence score.
Aware of the probabilistic nature of AI tools, the project team integrated a human supervision mechanism to trigger targeted human review when necessary. When the confidence score does not reach the defined thresholds, the contract is automatically referred for expert validation.
This approach avoids rigid workflows, strengthens user trust and progressively improves data quality.
CGI also played a key role in designing and delivering the proof of concept (POC) by developing a functional model—tested on real contracts—to demonstrate measurable business value.
The project highlighted the importance of prioritizing the workflow and interface over optimizing prompts in isolation, as well as the importance of close collaboration between IT and business teams to evolve the taxonomy in an agile manner.
The outcomes:
Concrete and measurable scaling
The results obtained during the POC were significant:
- Average processing time of a contract reduced from three hours to 45 minutes
- Extracted fields increased from 25 to 100 structured pieces of information per contract
The solution makes it possible to extract four times more structured information per contract while reducing processing time by 75%.
In addition, the platform proved particularly efficient from an operational standpoint, with monthly operating costs of less than €300.
Beyond these metrics, the transformation is structural. Contracts are no longer reviewed only manually; they now feed into a structured system, enabling more consistent, faster and integrated use of contractual data.
Airbus now has a unified semantic foundation that supports the use of real estate information at scale and strengthens the organization’s ability to manage its commitments effectively in a demanding, competitive environment.
Beyond the initial real estate scope, this semantic and technological foundation paves the way for extending automated document-processing capabilities to other areas of the organization.
With CGI, we have structured and improved the reliability of how we use our contracts on a global scale. AI, combined with an appropriate monitoring system, now enables us to transform our documents into data that can truly be used for management and decision-making.
Data Architect at Airbus
A long-term transformation
The project demonstrated that the industrialization of document AI depends as much on semantic structuring and governance as on the technology itself.
By combining business expertise, semantic structuring and supervised AI, Airbus and CGI transformed a fragmented volume of documents into a lasting lever for performance and competitiveness.
