Building with GenAI

Challenges, Learnings & a Bit of News from the Builders of Quivr.app

New features and learnings from the team at Quivr and news from across the exciting world of Generative AI.

Last week at Quivr

• Last week in Numbers

  • 4 Releases 🚀

  • 400 active users per day 👀

  • 6 contributors this week 🙏

  • 1300 new stars 🙌

• New Features and Improvements

  • Brain management -> You can now customize your brain!

Demo: Quivr.app - Source Code: GitHub

Event Announcement 🎉

Gen AI Days

The Global Gathering for Generative AI

Bringing together a vibrant community of forward-thinking professionals, AI enthusiasts, researchers, and industry pioneers.

GenAI Days is a series of meetups, conferences and publications with the goal of fostering collaboration, knowledge sharing, and innovation in the field of Generative Artificial Intelligence (GenAI) 👀

Check out the new site here [link]

Learnings from the week

• Stan - Sensei: Building a product is easy, a good product is hard. Adding features without bloat is the major challenge. Simple designs are often the hardest one to come to. Creating a new page for a new feature is easy, integrating is not so trival 🤷

• Zineb - Chief Dev: Remodeling Quivr as the vision changes creates its own problems. This week, we’re adding a file management system using supabase storage! 😎 Fun moment: When tech and product teams realize they’ve been discussing different features using the same components for the past week 😜

• Mamadou - Frontend Wizard: Good code quality is invaluable. Through gradual refinement of our frontend architecture, we now ship features with confidence, all thanks to rigorous testing 🧪 Our golden rule: communication via code - keep it simple, keep it readable 💫

• Matt - Backend Dude: Seperating the user from the contributor is super hard. When dealing with large LLMs, there are going to be some computations that need special hardware ⚙️ At some point we are going to break the golden rule of letting Quivr run for anyone 😬

• Chloe - Product Maestro: Creating a comprehensive product is cool, making it simple is essential! 3 steps that we followed with the Quivr team to achieve it: 1. Go back to basics and keep it simple. 😊 2. User experience has to be intuitive. 🤩 3. Challenge and Iterate. Question each feature and user journey 🌊

• Brian - Methodology King: Greatest products try to hide a maximum of complexity within the user interface. That’s our North Star at Quivr. Something is cooking very soon in that sense 😁

• Daniel - Feedback Warrior: Using the ARRR metrics, we found Quivr's retention below 2%, revealing a chance to reinvent the Quivr experience. We're making it simpler and more focused on what users want. Expect a reimagined, user-friendly Quivr soon - stay tuned! 😇

News from the World of Gen AI

• 🚨 BREAKING: Apple is building an AI chatbot called 'AppleGPT' to compete with ChatGPT. Apple is already testing the tool internally, and plans to launch it to consumers next year, according to Bloomberg.

• This article by Sergei Savvov on 7 Ways To Speed Up Inference of Your Hosted LLMs is very well written and is great for those looking to get into building with LLMs.

• Super Cool Open Source GenAI projects you should check out:

  1. RealChar by Shawn Wei

  2. Unofficial Code Interpreter API by Shroominic

  3. ShortGPT by Ray Ventura

  4. Petals by Big Science Workshop