Eki.Lab
Deep dive into Ekimetrics' centre of excellence for research and innovation.
Why an Eki.Lab at Ekimetrics?
- To continually improve our offers and our talents skills.
- To contribute to the development of the international scientific community, through the publication of scientific articles and open-source libraries.
- To open lines of research in collaboration with our clients on subjects at the cutting edge of technology, with high added value for them.
What is Eki.Lab?
- Eki.Lab is a structure that brings together over 50 PhD students, AI experts and Data Scientists.
- Eki.Lab is our centre of excellence. It allows us to keep a constant eye on the latest technical innovations, such as deep learning, MLOps industrialisation, responsible AI and NLP.
What are the research topics and how are they selected?
We have several types of research lines.
- In-depth: Time Series Modelling & Forecasting, Deep Learning & GenAI for NLP, Deep Learning & GenAI for Multimodality, Economy & Econometrics. These are background streams.
- Cross-functional: Optimisation, Explainability & Responsible AI, MLOps Industrialisation. For example, the project entitled “Detection and suppression of toxicity in a body of texts” is being carried out by the “Deep Learning & GenAI for NLP” and “Explainability & Responsible AI” teams.
How are the results integrated into business/client projects?
Our work on Gen AI began in 2021, focusing on multimodality and question Answering. This early start led to the release of Climate Q&A in early 2023, one of the first public RAGs, and in 2024, we introduced CLAIR, a multi-agent generative AI system. These innovations have enabled Ekimetrics to develop a rapidly deployable gen AI environment.
The integration of new algorithms into platforms such as One.Vision, our MMM solution. We emphasise explainability and causality in our AI systems, bridging business and science by delivering actionable insights. Our data scientists successfully apply their expertise to all client projects.
How is Eki.Lab different from other ones?
Multiple lines of research
As Marketing Measurement and Optimisation Leader (Forrester), we have opened lines of research on the following topics: Causality, Explainability, Demand Sensing & Time Series Modeling and Multimodality applied to Marketing.
AI applied to business
We have also opened up our own lines of research into all areas of AI applied to business. These lines are carried out in collaboration with customers or Tech partners such as Google.
Partnerships with Science
Finally, we are forging strong partnerships within the scientific ecosystem, including with INRIA, Lip6 and SCAI. Responsible AI, in particular, is at the heart of our research, since one of our missions, as a “Mission-Driven Company”, is to ensure that AI has a positive impact on the world.
What are the benefits of this Eki.Lab for our customers/talent?
- Cutting-edge solutions that incorporate the latest innovations in record time.
- A community integrated into a dynamic ecosystem (organisation of meet-ups, participation in scientific conferences, collaboration on projects with universities and Grandes Écoles, etc.).
Innovation path at Ekimetrics – Minimise time & maximise success
Illustration of our marketing solution from Eki.Lab to MMO Product OneVision®
- Uncertainty quantification with conformal prediction
- State-space model for customer journey modelling
- Benchmark coherence for code generation
- Monitoring of GenAI Pipelines
- Retrieve tabular data from chart
- Contextualised analysis of images for marketing
- Interpretability and customer centricity
- Image segmentation on MMO
- EBM for MMM
- Causality for Mix Marketing Optimisation
- Parsing of pdf document - EkiParser
- GenAI offer and industrialisation
- NLP and information extraction - Plume
- Long-term baseline extraction - Tresse
- Demand sensing - Horizon