SID takes care of the ingestion, interpretation and storage of the entire customer data infrastructure.
tl;dr
We allow you to connect your LLM apps to your customer's data in an afternoon using a single button and an "AI-friendly" API.
Ask: Please introduce us to companies building AI apps or considering adding AI to their product. We can save most teams months in engineering & MLOps.
❗Test the API with your own data: We have a free live demo at demo.sid.ai
Today, companies like Jasper spend months building integrations with their customer's salesforce, email, and docs. Previously, we made consumer AI apps (that went viral on Twitter) and found we spent 70% of our time doing data integrations instead of improving our product.
Why is it taking Jasper months?
There are two challenges, and developers tend to underestimate both. You must ingest the data (e.g., call Google's API) and interpret the data (i.e., go from chunks to insights).
Ingesting is a lot of grunt work: The APIs change, and a lot of infra is needed to keep information up-to-date – and deprecate the old. This is finicky and maintenance-heavy. And don't get us started on applying for Google Drive API keys: That can take months and costs ~$30k.
Interpretation is what's tricky. A naive embed + vector DB sounds compellingly easy but rarely works well for production-sized datasets. (If it does work for you, please reach out!) Despite what "Big Vector DB" is telling you, there is a lot of engineering and NLP that's needed: Recognize entities, model relationships, give it temporal awareness, filter the fluff, handle multimodal data, augment data with questions and summaries – I could go on. You must choose (& deploy) a language model for each of these tasks. And have you considered that the data distribution for Powerpoints is slightly but perniciously different than PDFs? Go directly to jail; do not pass "go."
Oh, and have you thought about compliance?
SID makes it easy for developers to connect LLM apps to their customer's data. We offer developers a button that lets people link their data and lets devs query our API to ask questions about that person or their company. That's it.
We take care of the ingestion, interpretation, and storage of the entire customer data stack – so devs don't need to think about vector DBs, maintaining infrastructure, or Google's API oddities. We self-host all LLMs used for analysis to ensure privacy – critical for customers. We've already served over a million API calls and average <200ms response times.
We've lived through the pain and built SID so future devs don't have to. We're starting with the most popular integrations and are adding new ones fast – at no extra cost to our customers. Please reach out if there's anything specific you need.
🇨🇭PS: We're Swiss, so we know how to keep your information secret.
We love being the odd ones: Lotte can fly a plane but can't drive a car, Georgijs still swears by Vim and built his own compiler at MIT, Lukas goes from idea to product faster than anyone the author knows, and Max is a researcher at heart who won't stop digging until he finds the truth – or people stop talking to him.