Add additional value to your data with the power of OpenAI embedded in your database. Easily build a Q&A bot to answer questions in your website, application, documentation or knowledge base.
Highly tuneable
All the knobs and dials you need to tune ChatGPT for your data
Speedy search
Xata provides native full-text or vector search that returns results fast
With a generous free tier
750K records and 15GB storage for free - OpenAI API calls included
You can import from CSV or use our Javascript, Typescript, or Python SDKs to populate the data. Our spreadsheet-like UI makes it easy to review and edit your database.
With a simple API call, ChatGPT will be answering questions using your data as context. You can provide your own rules to improve the results.
// Question a specific table. In this case "Docs"
const result = await xata.db.Docs.ask(
// Pass in a simple question
"How do I install Xata?",
{
// Rules define how your chatbot responds
rules: [
"Do not answer questions about pricing or the free tier. Respond that Xata has several options available, please check https://xata.io/pricing for more information.",
"When you give an example, this example must exist exactly in the context given.",
],
searchType: "keyword",
// Use search parameters to query and weight results
search: {
fuzziness: 2,
prefix: "phrase",
target: [
{ column: "title", weight: 4 },
"content",
{ column: "keywords", weight: 4 }
],
}
}
)
Tuning search is essential for getting the best results. The better the search, the better the context ChatGPT has for your questions. You can use our UI to tune the search experience by configuring column weights and value boosters.
If you are on the Xata free tier, you get 250 questions/month for free. You get 250 questions/month for each Xata unit you purchase. To see details on the Xata units, please visit our pricing page.
Yes, although currently not on our standard plans. Please contact us to discuss your needs.
Our examples repo contains a sample application that demonstrates using the API on several databases.
This YouTube video walks you through how we have used the Xata Ask API to add a chat-bot to our own documentation.