What is the Cape API?
The Cape API keeps your sensitive data private while using LLMs like GPT-4 and GPT-3.5-Turbo. Data privacy and security concerns form many barriers to adopting AI. Local LLMs are expensive to run, and maintain. OpenAI already has the best model and a low cost API, you just need to get it integrated while optimizing for privacy. With the Cape API you can easily upload sensitive documents, and start working with GPT-4: ask questions, summarize, output tables, compare files, and whatever else you can think of. The data remains private.
We encourage you to try out CapeChat, which is built atop the Cape API.
How does it work?
The Cape API leverages confidential computing for end-to-end secure processing. AWS Nitro Enclaves are secure, isolated, cryptographically verifiable virtual machines purpose-built for processing highly sensitive data. Within the enclaves we confidentially perform entity redaction and manage an encrypted vector storage and search service. When sending a prompt to GPT-4, OpenAI and their sub-processors can only ever see the redacted data.
The API exposes granular control of essential privacy tools including de-identification and re-identification, supporting redaction for 50 types of entities like names, emails, credit cards, addresses, and other PII/PCI/PHI across 47 languages.
How do I get started?
- Create a Cape account by signing up for CapeChat.
- Enable the Cape API on the billing page (must enter a payment method).
- You're now ready to generate an API key and start calling endpoints!
- Follow this tutorial to use the Cape API to de-identify text or files.
- Follow this tutorial to use the redacted data with OpenAI.
Join the community
Join our Cape Community Discord to ask questions, get answers, and hang out with other privacy-minded developers.
Learn more about Cape’s implementation of confidential computing.