Stan Girard Profile picture
πŸ‡«πŸ‡· @quivr_brain (@ycombinator W24) πŸ‘‰ https://t.co/Qx8QYiemol
2 subscribers
Nov 14, 2023 β€’ 7 tweets β€’ 2 min read
I spent a few days understanding OpenAI value propositions so that I can create a product that scales with them and isn't just a wrapper πŸ‘€

How can you create products in a market where one company dominates and iterates so quickly ? 1️⃣ Build a product that scales with OpenAI

I've marked with a βœ… issues and features where I think OpenAI will increase/fix or implement features. If you product's features scale with the improvements of these issues, then you are on track.
Aug 2, 2023 β€’ 15 tweets β€’ 5 min read
Spice up your AI game with Genoss GPT! πŸŒˆπŸ‚

Imagine swapping @OpenAI with any LLM, all in a single line.

- Switch to LLama V2 from @Meta? Done
- Experiment with GPT4ALL from @nomic_ai? Easy
- Custom model on @huggingface? Absolutely

Made w/ @LangChainAI & @huggingface

πŸ§΅πŸ‘‡ Image @OpenAI @Meta @nomic_ai @huggingface @LangChainAI Genoss GPT is an API that is compatible with @OpenAI SDK

It means you can replace OpenAI GPT3.5/4 url with Genoss URL and it will work.

Here on the @streamlit demo you can either use:
1 - @OpenAI through their URL
2 - Or via Genoss Image
Jun 26, 2023 β€’ 7 tweets β€’ 3 min read
OpenAI Functions can make your app 10 times better! πŸ”₯

But you can easily use them wrong.

Here are 5 mistakes I made and how I fixed them on @quivr_brain @quivr_brain Limit the number of functions as their description counts as tokens.

@quivr_brain functions represent 341 tokens each call

You can count your tokens here https://t.co/bTv9IWTdq8platform.openai.com/tokenizer
Jun 24, 2023 β€’ 11 tweets β€’ 3 min read
πŸ”₯ OpenAI Functions made @quivr_brain 5 times faster πŸ”₯

If you are not using it, you are missing something. It is a game changer.

Here is how it works 🧡 @quivr_brain We created a basic prompt that gives @quivr_brain its goals and the functions it has access to
Jun 14, 2023 β€’ 5 tweets β€’ 4 min read
πŸ”₯ New @quivr_brain release πŸ”₯

- GPT3.5/4 compatibility
- API Key management system

You can now have 16k context in Quivr AND generate your own API keys to create your own tools on top of it πŸŽ‰

Thanks @LangChainAI & especially @hwchase17 for the quick work !

🧡 1 - API Key

You can now create your own API keys to interact with api.quivr.app from anywhere πŸ”₯

More info on brain.quivr.app/docs/backend/a…

@mattzcarey
Jun 12, 2023 β€’ 17 tweets β€’ 45 min read
Your GenerativeAI Second Brain is 1 month old! πŸ”₯🧠

What a trip you guys! Thanks for the support! 7k ✨ & 1k users 🀯

One Limited Edition T-shirt to win!
RT / Like to get a chance

The Story and Future of @quivr_brain in πŸ§΅πŸ‘‡ Image @quivr_brain Code with chatGPT4

Two months ago I started working on understanding what could be done with GenerativeAI.

I made my chatGPT Web UI alternative. Entirely coded with GPT4 🀯

Source code: github.com/StanGirard/Cha… Image
Jun 6, 2023 β€’ 11 tweets β€’ 13 min read
How do you use chatGPT to chat with your docs?

Here is how you can chat with PDFs, Powerpoints, Excels, etc ...

🧡
[save this] Image 1 - Ingest the Documents with @LangChainAI

You can use one of the many Langchains document loaders to load any type of document.

From CSV to PDFs, it has you covered.
In @Quivr_ap we use the PyMuPDFLoader that extract the content from the PDF Image
Jun 4, 2023 β€’ 8 tweets β€’ 4 min read
Quivr's new web app is now in private beta! πŸ”₯

To get access simply like and comment what will be the first question you will ask your Generative AI second brain

All of this is powered by @LangChainAI, @supabase & @OpenAI Image Make sure to follow me so I can send you a message πŸ˜™
Jun 3, 2023 β€’ 9 tweets β€’ 6 min read
How do you make @supabase, @LangChainAI, @OpenAI, and @AnthropicAI work together to build an app such as @Quivr_app ?

Process explained in 🧡 @supabase is super easy to integrate into @nextjs

supabase.com/docs/guides/au… Image
May 15, 2023 β€’ 4 tweets β€’ 3 min read
Quivr is trending on @github ! πŸ”₯

Thanks for the support !

The repo is only two days old πŸ”₯🀯

Lets make an awesome second brain powered by GenerativeAi thanks to @OpenAI @supabase @LangChainAI , @streamlit

Link in 🧡 Image github.com/StanGirard/qui…
May 14, 2023 β€’ 6 tweets β€’ 4 min read
Quivr.app - Your GenerativeAI Second Brain 🧠 is getting @pinecone Vector Store support very soon!

Anybody will be able to add the vector store of its choosing!

- @trychroma
- @qdrant_engine
- @elastic
- @supabase
- @weaviate_io
- @activeloopai Deep Lake

🧡 Image First I created a Vector Store Interface

- Select: allows the user to retrieve a vector
- Delete: allows the user to delete one

It allows anybody to add a vector store by implementing these two functions. Image
May 14, 2023 β€’ 4 tweets β€’ 2 min read
Quivr - The Obsidian GenerativeAI replacement just got a website πŸ”₯

Not much, just learned @vercel and @nextjs in the last 2 hours but it is a start πŸ˜™

Thanks for the support ! The github is already 250 ✨ strong !

Link in 🧡 Image 1 - Github

github.com/StanGirard/qui…
May 12, 2023 β€’ 10 tweets β€’ 7 min read
GenerativeAI as your second brain ?🧠

Dump all your knowledge into a vector store with any format ! Audio, URLs, text, code.

Query it with GPT4.

Created to replace my @obsdmd Vault using @streamlit , @LangChainAI and @supabase.

🧡 for more details Image For example, the markdown loader is called when it is a markdown file.

1 - Load the markdown with @LangChainAI
2 - Calculate the SHA1 of the file
3 - Add the SHA1 as Metadata
4 - Add to the @supabase VectorStore

Next time, if the file has the same SHA1 it won't be uploaded 😎 Image
May 7, 2023 β€’ 6 tweets β€’ 1 min read
How to create beautiful images with Midjourney with the help of GPT4 ?

Here is a prompt for you:

I want you to act as a prompt engineer. You will help me write prompts for an ai art generator called Midjourney.

πŸ”₯πŸ‘‡ Image I will provide you with short content ideas and your job is to elaborate these into full, explicit, coherent prompts.

Prompts involve describing the content and style of images in concise accurate language.

πŸ”₯πŸ‘‡