bit.io

bit.io is joining @databricks ! We are sunsetting our database product on June 29th. Check out our blog for more. Thank you to all of our users. blog.bit.io/whats-next-for…

Good piece by @bitdotioinc on why we should care about vector databases. innerjoin.bit.io/why-you-should… h/t @WriteAheadLog @JMoMort

We're better at releases than you: blog.bit.io/we-are-better-… Iteration speed & developer productivity is the biggest advantage of a startup. @bitdotioinc is 8 devs and in the last few months we've shipped: multi-region & multi-cloud, service accounts, text->SQL, new pricing, performance & reliability improvements, mobile support, and more.

Try it for free on bit.io. In our latest post in The Inner Join, we guide you through the process of getting vector embeddings from your documentation and conducting semantic search with bit.io+pgvector. innerjoin.bit.io/vector-similar…

No need for specialized vector databases—with #Postgres and bit.io, integrate semantic search easily into your projects. Store vector embeddings and perform searches with the pgvector extension—all within your Postgres database.

Vector search enables powerful semantic search capabilities, for searching by meaning rather than keyword. It uses large language models, embeddings, and similarity metrics. Now, you can implement it in #Postgres using the pgvector extension. innerjoin.bit.io/vector-similar…

Great to see this coming together of the #PostgreSQL and #mysql community.

In our latest post in The Inner Join, we show how to use the pgvector Postgres extension on bit.io for vector similarity search—all in #PostgreSQL, no other specialized tools or integrations needed. innerjoin.bit.io/vector-similar…

@alexgraveley Yep! I think we (bit.io) are by far the least expensive & fastest way to use pgvector with Postgres.

ChatGPT writing SQL innerjoin.bit.io/make-chatgpt-s… #sql

@Exploringfornow Ouch ☠️ @bitdotioinc’s postgres database with pgvector works out much much cheaper at $5/month after a super generous free tier

@yar_vol @alexgraveley @CockroachDB bit.io supports pgvector!

@alexgraveley Neither @bitdotioinc (bit.io serverless Postgres) nor @CockroachDB (proper cloud database) support this, so we have to rely on traditional sorted indexes and serialising vectors into sortable strings (KD partitioning)

@GoogleColonizer @danjliden That said: the last thing we want is for the model to confidently answer an ambiguous question. Being able to flag ambiguities or at least provide an interpretation of the prompt before answering will be hugely beneficial.

@GoogleColonizer @danjliden In our limited testing of ambiguous prompts, LLMs have been pretty good at detecting ambiguities and can, if told to do so, ask for clarification. But users still have to ask the right questions. While LLMs can assist with SQL, they can't always rectify flawed logic.

In our latest post on LLM text-to-SQL translation, we discuss how to test and compare task prompts/prompt templates—that is, how to make sure that well-constructed user prompts consistently lead to working SQL. We then compared #ChatGPT to Codex models. innerjoin.bit.io/llm-sql-transl…

LLMs, when integrated with other tools, unlock a world of potential applications beyond end-user text delivery. In this post, we provide some examples of how integrations with other tools can enhance text-to-#SQL translation. innerjoin.bit.io/use-llms-with-…

Are you a #NodeJS developer? Take a look at our Node.js SDK for connecting to bit.io databases and interacting with the bit.io developer API. @bitdotioinc/node-bitdotio" target="_blank" rel="nofollow noopener">npmjs.com/package/@bitdo…

This is part of a series on our efforts to make the best production-ready text-to-SQL translation system. Keep an eye on this feed for all the updates: innerjoin.bit.io/ai/home

We wrote about the process of switching from the Codex API to the ChatGPT API here: innerjoin.bit.io/make-chatgpt-s…

With @OpenAI discontinuing the Codex API, you might wonder what's going to happen to our #AI text-to-SQL functionality, which is based on Codex. We'll be switching the system over to the #ChatGPT API in the next couple of days—you shouldn't notice any difference or interruption!

Now that @OpenAI is discontinuing its Codex API, you might want to switch to #ChatGPT for code completion. The chat-based API calls are different, but you can still get very good code completions. Here's one way to do it! innerjoin.bit.io/make-chatgpt-s…