Nuclia, the RAG-as-a-Service platform
Nuclia automatically delivers AI search and Generative Answers on top of your unstructured data, and provides knowledge in the form of trusted answers, in any language, and from any data.
Nuclia allows you to customize your RAG pipeline and bring it directly to production. With Nuclia you can:
- Index any type of data in any language.
- Define your own retrieval strategy.
- Use any LLM to gain insights and knowledge from your data.
- Customize your prompt.
Last important updates
🚀 LLM & Model Enhancements! 🤩
We're excited to announce some fantastic additions to our model lineup:
Llama3.2 Joins the Family: Give a warm welcome to our newest model member! Llama2 is here to bring its A-game to your AI workflows.
Anthropic's Latest and Greatest: We've added the new Claude 3.7 and Claude 3.5 Sonnet v2 to our roster.
Azure Mistral Large 2 - The New Hotness: Heads up, team! We're saying goodbye to the old Mistral model. Starting Monday, Azure Mistral Large 2 will be your go-to model. Out with the old, in with the new!
🛠 API Updates: Powering Up Your Development
We've been busy making our APIs more robust and flexible:
The /ask endpoint is getting supercharged! 💪
- Image Upload Magic: You can now pass images to the /ask endpoint using extra_context_images. Visual context? Check! 📸 say cheese and just remember you need you use a visual LLM for answering
- Search Configuration Simplification: Customize your API search directly from the dashboard by passing the search_configuration name. Less coding, more doing! 🚀
Vectorset Migration—sounds fancy, right?
It simply means you can now add multiple embedding models to your Knowledge Base and switch between them with ease. The /ask endpoint handles all the heavy lifting.
Advanced Filtering: We've supercharged the filter_expression parameter for /find, /ask, and /catalog endpoints:
- More granular control over field and paragraph filtering
- /catalog now supports searching over origin metadata during retrieval
Enterprise-Level Backup
Enterprise account holders can now create Backup/Restore knowledge boxes. Data safety first! 🔒
💅 Agents Front
Calling all current clients! Our Ingestion Agents are ready for their beta debut:
- Labeler: Automate your classification processes like a boss.
- Generator: Create summaries, JSON outputs, and extract structured content.
- Graph Extraction: Enrich resources by extracting Named Entities and relations.
- Q&A Generator: Build your own resource-specific FAQ.
- Content Safety & LLM Security: Flags inappropriate content faster than you can say "oops!"
Bonus: You can now trigger agents based on labels. Workflow automation just got a serious upgrade! 🤖
🔍 Quality Tracking & Visibility
We believe in transparency:
- Remi logs are now downloadable for quality monitoring.
- Activity logs are now visible in the UI for better tracking!
See the full change log.