We’re @sumants and @rmehtany, working on Pontus.
Pontus makes it easy to use AI with privacy embedded.
We were concerned about the volume of personal data that goes to large LLM models without protection. We tried find an easy solution where didn’t change the simple apis given by LLM providers. However, most required you to invest significant engineering effort.
We wanted privacy and LLMs to be easy, so we built Pontus. Through a declarative YAML, we orchestrate a microservice with the most common element of the LLM stack.
- Anonymize Prompts before it hits LLMs, yet keeps context on your servers - Secure RAG that embeds documents safely - Semantic Cache that doesn’t store PII at Rest
We did this not only because it is the right thing to do, but the regulation landscape makes it critical to build with privacy first. Recently, many companies have been fined and forced to delete their AI models [1][2]
For us, Privacy is by design, not an afterthought.
Please visit us here
Our Open Core Repo: https://github.com/PontusAI/Pontus Our Website: https://pontus.so
[1] https://www.ftc.gov/news-events/news/press-releases/2023/05/...
[2] https://techcrunch.com/2023/05/10/clearview-ai-another-cnil-...
