🚀 Big News! PrivateGPT expands its horizons with the introduction of Zylon, the AI collaborator for every workplace.
Zylon emerges from stealth with a $3.2M Pre-Seed led by
@Felicis
to revolutionize GenAI usability and privacy.
Join the beta at
PrivateGPT is a production-ready service offering Contextual Generative AI primitives like document ingestion and contextual completions through a new API that extends OpenAI’s standard.
Talk to your documents privately using the default UI and RAG pipeline or integrate your own.
In case you missed today's
@llama_index
webinar "Building Production RAG with Local Models (featuring PrivateGPT)", here you can find the slides we presented with a complete overview of the project so far 👀
Thanks to
@jerryjliu0
for hosting the team 🙏
🚀 Exciting Project Update: v.0.2.0 is out!🚀
Contributors have been working hard transforming PrivateGPT for a faster, more secure, and customizable experience! 🌐 Dive into the details below 👇
@llama_index
has been a reference project for us since day 1, so we are proud to be speaking about PrivateGPT, RAG design, privacy in GenAI products and much more in their next webinar 😊
📺 Join us and
@jerryjliu0
on Wednesday at 9am PT
Want to learn how to build a production-ready RAG stack using purely local models?
We’re excited to be hosting a special webinar with
@ivanmartit
and
@thanos_malkav
on
@PrivateGPT_AI
- the leading project for privacy-aware RAG.
it’s powered by
@llama_index
, and you can also
Less than a month after releasing our new revamped version, PrivateGPT is back to being the Top Trending repository on
@github
🥇 while passing the 40K stars ⭐️
THANKS everyone for your support!!! ♥️
Stick around, great things are coming to this project and its community 🚀
Really interesting way of bringing the power of ColBERT () to use advance retrieval techniques based on fine-grained contextual late interaction 👀
Should we create a new RAG implementation in PrivateGPT following this approach❓
The RAG wave is here to stay, but in practice, it's hard to retrieve the right docs w/ embdings, & better IR models are hard to use!
Let's fix that: Introducing 🪤RAGatouille, a lib to train&use SotA retrieval model, ColBERT, in just a few lines of code!
Happy to announce that
@qdrant_engine
has become the default
#VectorDatabase
for us 🚀
As a result, we incorporate high-performance vector similarity search into
#private
AI applications 💪
Cheers to a fantastic collaboration! 👏
👀 Integrations docs:
📺
@MatthewBerman
just published a new great video describing all the new features we have recently launched
🙏Thanks again for supporting us since day 1 and helping sharing the project with the community
🎁 Stay until the end for a little surprise!
📃 Sources in Completions API:
Get info (text + metadata) about the Chunks used during inference by the contextual Completions APIs by adding:
{"include_sources": true }
to the request Body.
Sources are enabled by default in the UI.
👩💻 Docs:
🧠PrivateGPT's Ingest API returns the *doc_id* and the extracted *doc_metadata* - btw, it works great with
@llama_index
's Metadata Extractors!
Here are a few cool use cases that illustrate why this is incredibly useful for developers: 🧵
Here is a great hands-on article by
@CognitiveTech_
who uses
@PrivateGPT_AI
to compare the impact of
- 🤖 Models
- 🔡 Prompt styles
- 🦾 System prompts
- ✍️ User prompts
on private, AI-powered book summarization
Article:
Project:
💬 And for those just wanting to talk to their documents in a plug-and-play way, we have made things easier with the new PrivateGPT UI!
Now you can upload documents, select the interaction mode with your LLM (query, chat or chunks) and talk to it through a prompt.
🗓️ 3 months ago we launched the brand new version of PGPT along with its Discord server and we are already 2.3K members collaborating on a daily basis in it:
Big thanks to this amazing community: you are the real fuel of this project 🥰🚀
🚀 After 38K+ GitHub stars since its creation in May 2023 and all the contributions (5K+ forks) and feedback provided by the open source community, we have re-designed PrivateGPT from scratch, focusing on value added and software quality.
👨💻
@ivanmartit
(PGPT co-founder) will collaborate in the
#hackathon
organized by
@GoHubVentures
to develop
#AI
apps where data privacy play a center role.
🇪🇸 If you can visit Valencia (Spain) on January 25, don't miss the chance to come by and say "hola"
Check out our new Context Chunks API:
🔎Given a text, returns the most related chunks from the ingested documents.
➕Includes previous and next chunks for extended context retrieval.
⏱️Uses only the embeddings model, so it is hyper fast!
Docs:
💡We have extended OpenAI’s API to add support to document-based contextualised completions, abstracting the complexity of RAG pipelines.
It is as easy as including "use_context" attribute to your request:
💻 If you want to go deeper, our architecture allows you to customize every step of the RAG pipeline and the different components involved: embedding model, LLM, vector data base… letting you to execute locally (even without Internet connection) and on a private cloud.
@mozilla
❤️
@PrivateGPT_AI
Mozilla Innovation is using PrivateGPT on a very interesting innovation project, take a look!
- Hacker News front-page:
- Project announcement:
- Project site:
Zylon is the first end-to-end Collaborative GenAI tool in the market, allowing companies to gather their teams to work together in shared projects designed to foster collaboration and creativity around different company data sources.
This wouldn’t have been possible without the support of our lead investor
@felicis
, as well as
@LifeXvc
and
@zypsycom
who trusted our vision since the early days of PGPT. Read the full story published today by
@VentureBeat
written by
@carlfranzen
Zylon is a B2B SaaS powered by PGTP that provides features like summarization, comparison, report generation and insights exploration over different data sources with no tech expertise or prompt engineering skills required, leveraging a novel non-chat-based user experience.
If you want to be up to date with all the news related to Zylon, we have created a newsletter that, alongside our blog, will allow us to be in touch and communicate all the amazing features to come, as well as discussing about AI in the workplace:
Zylon is the consequence of all the learnings we have achieved with PGPT in the last 9 months. We plan to evolve both projects together by incorporating Zylon’s enhancements into PGPT. Hence, this account will cover the whole family from now on!
Privacy has been considered since inception by leveraging PrivateGPT tech, thus allowing customers to have full control over their data in a SOC2-certified SaaS environment with the possibility of a full on-premise setup for enterprises that require extreme privacy.
@fipra
No problem, we totally got you 😉
To delete documents you can use the recently added API . But be aware that we haven't included a button to use it yet in the UI.
Join our Discord to follow the discussion about it
Zylon brings a unique interaction in the GenAI field by understanding the “what, when and why” behind the needs of every business to proactively execute tasks without explicit user input, expanding the frontiers of what customers can expect.