An AI data analysis platform integrated into the SpinalCom BOS
The Building Operating System (BOS) developed by SpinalCom is not limited to the aggregation of contextualized data. It also includes an Analytics module, enabling the creation of business calculation models directly based on the building’s contextualized data.
The results produced are stored and can then be leveraged through several channels:
- Integration into a BI tool (Power BI, Tableau, etc.)
- Custom development of dashboards integrated into the SpinalWall portal, ensuring consistent ergonomics with other SpinalCom modules
- Automatic synchronization with other applications
AI data analysis: raw data is not enough
In the smart building world, collecting data does not guarantee its value. To generate a relevant analytic, each data point must be contextualized: knowing which building, floor, room or equipment it corresponds to, what function it fulfills, how frequently it changes, with which other data it is correlated, etc.
Without this contextualization, analysis models are unusable, and artificial intelligent projects fail quickly due to lack of structure. This is precisely what SpinalCom’s architecture resolves, by merging physical asset description, real-time (OT) and transactional (IT) data in a unified graph database.
Graph database vs relational database: architecture makes the difference
Many solutions rely on relational databases where measurement points are simply “tagged”. This is like labeling thousands of post-it without knowing the overall logic. SpinalCom’s graph database, on the other hand, structures the relationships between each data point (location, function, nature, history…), which allows:
- Intelligently navigating the graph to create complex analyses,
- Guaranteeing the scalability of deployed AI models while providing them with quality data,
- Avoiding calculation errors or redundancies.
Concrete analogy
Relational database with tagging : an Excel spreadsheet with no tabs or formulas, just filled columns.
SpinalCom graph database : a data graph, in which each sensor or equipment is linked to all its context and history
Relational Database vs Spinalcom graph database
Comparison of data management approaches
| Criterion | Relational database with tagging | SpinalCom unified graph database |
|---|---|---|
| Structuration | Flat tables + tag fields | Explicit relationships between objects and events |
| Spatial contextualization | Manual tags “building”, “room”, etc. | Natively integrated via the graph hierarchy |
| Business understanding | Hard to read without documentation | Intuitive navigation by relationships (e.g.: sensor → room → floor) |
| Maintenance / scalability | Growing complexity as the system evolves, difficult to replicate | Dynamically adaptable, scales without breaking the model, industrialization facilitated |
| OT/IT/Real estate interoperability | Siloed by tables or data sources | Merged in a unified structure. Shortcut access to information through relationships |
| Analytics / AI readiness | Requires significant quality improvement and mapping work | Natively contextualized data, ready for analytics / AI use |
| Analytics implementation time | Long and unstable | Short and reliable thanks to the common model |
| Typical use case | Static reporting, simple dashboards | AI, predictive maintenance, load shedding, smart control |
According to Gartner, knowledge graphs or contextualized graph databases have become a central element of modern data architectures, as they provide the context and semantics essential for the success of AI projects. Unlike traditional relational databases, they allow data to be linked intelligently, reliable insights to be extracted, and algorithms to be made more explainable, robust and scalable.
Concrete benefits for your AI data analysis projects
- Rapid deployment of custom analytics from building data.
- Smooth integration with your BI tools or internal dashboards
- AI readiness through a clean and contextualized database
- Creation of predictive maintenance models, automated energy load-shedding… directly usable and reusable.
- Native interoperability with all SpinalWall modules
Key figures on AI data analysis
80 %
of time spent on an AI project is dedicated to data structuring (source: McKinsey)
85 %
of AI projects fail at scale due to lack of a contextualized database (source: Gartner)
Up tp30 %
reduction in maintenance costs through predictive AI models (source: Deloitte)
20 to 25 %
energy consumption avoided through real-time automated load shedding (source: IEA Smart Buildings 2024)
An ambitious AI roadmap, with concrete cases already in production
The SpinalCom platform is already used in predictive maintenance projects, analyzing anomaly and failure histories to anticipate critical interventions. Maintenance then becomes proactive by anticipating problems before they occur.
Our AI roadmap also includes:
- Adaptive energy management, based on occupancy and energy mix,
- Automated load-shedding algorithms, to optimize consumption and power demand
- Real-time anomaly detection, for sensitive environments such as hospitals, airports, data centers…
Discover the potential of AI data analysis for your buildings Would you like to structure your data to succeed in your analysis and artificial intelligence projects?
Request a demo of the SpinalCom platformThey trust us
