Debkanchan

https://github.com/debkanchan/sdking

Hi all, I'm DK.I've built a few products before and every time I hated writing fragile untyped fetch requests for the ever-changing backend REST apis. I tried to solve it by using tRPC but tRPC caused build pipelines to become unnecessarily heavy and complicated. Also, tRPC/gRPC might not be suitable for everyone and REST is widely used.Rn we are building a react project and want to tear out...

https://platform.openai.com/docs/api-reference/responses/object

My requirement is simple: - folder based routing like next.js. - route loaders for SSR - not a pain in the buttWhat we tried: Remix/ React Router: - folder based routing doesn't work. /cheese/cake -> cheese.cake/route.tsxTanstack Start: - Too new. Don't have confidence yet.We are currently on Next.js but app router doesn't have getServerSideProps and generally it's a big pain in the butt to...

Right now I am experimenting with various parsing and chunking strategies for our LLM application. We are using self hosted ChromaDB.It feels odd to create a collection for each strategy to experiment with. It still works in dev but in prod creating a new collection every time with the relatively big amount of data (35k PDFs which translate into ~600K Docs of ~2000 character length) and then...

Full Disclosure: I'm the author of genkitx-graph and been part of Firebase Genkit early access. Still i'm using langchain & langgraph because they are the industry standard and have more functionality compared to genkit. But, compared to Firebase Genkit it has a very primitive DevEx1. There's no Dev UI to play with chains, runnables and models.2. There's no local tracing for debugging.3....

I am trying to figure out how to securely share and sync .env files and secrets with my team. Is there a saas/oss solution that "just works" or there's no way around using hashicorp vault/GCP Secrets/etc. with custom scripts?

I want to rapidly iterate generated embeddings. ChromaDB has collections so I can create a new collection for every iteration/version/commit but in my application I need to a) easily query the latest version of embeddings preferably without having to change the collection name string manually every time b) still be able to manually select a version (for A/B test) and c) quickly rollback...

https://www.youtube.com/watch?v=jLVl5V8roMU

https://www.npmjs.com/package/genkitx-graph

Firebase Genkit provides a really featured dev UI/playground for testing flows(chains for Genkit), prompts, tools and models and a dev trace store. Is there something similar for langchain?

https://mbleigh.dev/posts/ai-data-filter/

https://paralegal-ai.vercel.app/

https://blog.rideapp.in/conquering-errors-in-go-a-guide-to-returns-and-handling-a13885905433?gi=3af77c567f4b

I love neon.tech but the only problem I have committing to it rn is that it's only available on AWS and my entire infra is on GCP and I do not wanna pay data transfer fees.

I saw some Secret Manager charges and I was damn sure I was under free limit (6 secrets/mo).I found out whenever you create a Connection in Cloud Build, It creates an OAuth token in the secret manager.If you're just testing cloud build, be sure to cleanup secrets for deleted connections.