DBT team submission

README.md

@@ -14,5 +14,12 @@ From **June 2–4, 2025**, join us at **Campus Jourdan (Paris)** for a three-day
 
 It’s your turn — time to build your team, explore challenges, show your skills and shine.
 
+## ⏳ Before HackDays: Connect, Create, Collaborate!
+
+- [Explore existing ideas](../../issues) or [propose your own](../../issues/new) — every idea is welcome, big or small!
+- Find teammates, discuss projects, and brainstorm together in [discussions](../../discussions).
+- When you're ready, join a team or create a new one by making a [pull request editing](teams.md).
+
+### Happy hacking! 🚀
 ## 📄 [Event documentation](https://impress-staging.beta.numerique.gouv.fr/docs/94003084-ef31-4272-876e-ce25577ef804/)
-For you participants! (Progam, WIFI, Tech Docs etc.)
+For you participants! (Progam, WIFI, Tech Docs etc.)

submissions/dbt/README.md

@@ -4,27 +4,37 @@
 FactVerifAI
 
 ## Project Description
-Add fact checking to Docs with AI.
+FactVerifAI adds seamless, AI-powered fact-checking capabilities to Docs, an open-source collaborative document editor. Users can highlight text and receive a real-time accuracy score backed by sources and transparent reasoning. The goal: empower users to write and collaborate with greater trust in the facts they share.
+
 
 
 ## Contributors
 <a href="https://github.com/a-i-slo">@a-i-slo</a>, <a href="https://github.com/lluli">@lluli</a>, <a href="https://github.com/nboyse">@nboyse</a>, <a href="https://github.com/dalebird">@dalebird</a>, <a href="https://github.com/developwithtom">@developwithtom</a>
 
 ## Code base
-(Provide a link to your Git repository)
+https://github.com/lluli/docs 
 
 ## Deliverables 
 (Provide a link to a live demo, if you have one)
 (Add screenshots (image, gif or video) and presentation deck to `/assets`)
 
 ## Key Achievements
-(Highlight the main features or breakthroughs)
+Built a package using Python which supports locally hosted LLMs via Ollama and Open AI backends
+Real-time factual scoring system with source transparency (links + explanation)
+Collaborated with DINUM and La Suite to align with European digital sovereignty goals
 
 ## Challenges Overcome
-(What was difficult? What did you solve?)
+Balancing speed and accuracy in the fact-checking model
+Ensuring reliable source extraction and scoring within limited time constraints
+Integrating seamlessly into the Docs platform while respecting data protection and transparency principles
+Managing multiple language/model dependencies and preparing for multilingual use cases
 
 ## Impact
-(Who will benefit from your project?)
+FactVerifAI supports anyone working in collaborative document environments by enhancing trust and accountability. It contributes to European digital sovereignty by offering a sovereign, open-source alternative to fact checking solutions.
 
 ## Next Steps
-(If you were to continue, what next steps or developments would you envision?)
+Expand support for multiple languages
+Optimise performance and scalability for wider public deployment
+Adding more LLM backends such as Agno Framework
+Integrating the feature with a RAG system and allowing users to fact check full documents
+Conduct user testing with public sector teams across the EU to refine UX and functionality