Hello @everyone

Hi! Quick question about Nexus. I know the managed sandbox is preview-only and GA will be BYOC. Is Nexus BYOC enterprise-only, or is there a path for smaller teams? And any rough pricing guidance yet (Pinecone fee plus cloud infra)? Trying to figure out if it's feasible for a smaller setup. Thanks!

I have extensive experience with RAG and pinecone.

amazing to hear. what are you building these days?

I am building AI agent platform to help business. And I am looking for AI project that I can help in free time, too.

Hey! As we get closer to GA with Nexus, we're looking for a few community members to share their real, first-hand experience — a LinkedIn/X post, video, or whatever your format of choice is. No pressure on polish, just your honest take on getting started, what's worked, or where you'd like to see Nexus go next. If you've been using Nexus and are up for sharing, DM me and let's chat!

Pinecone Nexus is GA ➡️ https://discord.com/channels/1383658435366621315/1408122238837985340/1534933801682206721

Good

Hey all! This is Arjun from the Pinecone DevRel team. I'm looking to learn more about how people ask their coding agents about new products and tools to try. I'm curious, has anyone here asked Claude Code about a new devtool, and it did a websearch for you? Or, do you ask for websearch when considering an approach explicitly? Or, do you just let your coding agents tell you some tools to use and start there? Thanks in advance!

👋 I’m currently developing Agelgay AI, an AI-powered Ethiopia Service Guide, and Tebeqa AI, a legal guide and assistant focused on Ethiopian laws and legal documents. I’m using Pinecone for vector search/RAG, and I’m currently working on indexing a large collection of Ethiopian legal documents. I ran into a strange authentication problem today. Pinecone was working normally earlier and successfully uploaded around 1,000 chunks/vectors. Then it suddenly started returning:

401 Invalid API key

I generated a completely new API key, but the problem continued. I also tested outside the Python SDK using:

curl.exe https://api.pinecone.io/indexes -H "Api-Key: $env:PINECONE_API_KEY"

and it returns:

Invalid API key

I have confirmed that my environment is reading the new key correctly, with no extra spaces or characters. My account is active, the status page is green, and I only have one Pinecone project. I’m hoping someone here can help me identify what might be happening and get Pinecone working again. 🙏 Looking forward to learning from everyone here!

Can you verify that you're using the right API key for the organization and project that the index is in? This sometimes happens - a mismatch and it's in a different org/project. https://app.pinecone.io/organizations/-/projects/-/keys

Is there any roadmap/timeline on Canada Regions for Pinecone?

🧵 thread for sharing projects for the community webinar next month!

Awesome

Hi all. I would like to share my latest article. I am writing a series of articles covering AI Engineering Interview questions. It will go from easy to hard in the span of 5-7 articles. Please give it a read. 😊 https://medium.com/@nitanshuj138/ai-engineer-interview-questions-level-1-8e95789c392f

Looking for teammates who needs a part-time job

Hey @everyone ! We launched pg_pinecone and are working on fun demo/ideas for it. Here's one I'm making, I'm looking for testers/feedback. Of course, if you have any questions or ideas yourself, join the pg_pinecone channel and let us know 🙂 Here's what I'm making: A tiny local docs-search tool that lets you hit any docs site's llm.txt and build a search engine for your coding agents. Super handy for customizing the context your agents should have https://discord.com/channels/1383658435366621315/1564394337155027055/1565779817042022542

My attorneys will be reaching out to you

For copyrighted material of my pinecone methodologies

Can you send me more information

@everyone We're live and hanging out! Come through with questions and comments! https://www.youtube.com/@pinecone-io

Hey

hello, great to meet you! What brings you to the discord? 🙂

What chunk size are you running on your docs, and have you checked how many of your queries can't reach their answer page at all, even at k=10 exact scan? I found 7% in a Python docs index and it turned out to be chunk boundaries, not the index.

that's a neat way to catch chunking issues. did that 7% mostly come from long doc pages, or short API refs where the relevant bits get split apart?

A bit of both, but for different geometric reasons. On long docs, standard chunking breaks up continuous context windows splitting cause from effect across arbitrary token boundaries and causing the model to hallucinate missing connections. On short API refs, it’s a density problem: when a 2-line code example or parameter definition gets sliced in half at a chunk edge, the functional payload loses its structural integrity completely. In our testing, the 7% retrieval recovery came from replacing fixed-token slicing with dynamic boundary checks treating the document as a continuous manifold so related parameters and context stay anchored together regardless of page length.

that makes sense, especially on short refs where one split example can lose its whole meaning. how are you deciding boundaries in practice, based on headings and code blocks or some other signal?

Hi there, I am looking for long-term collaboration partner.

structural signals (headers + AST code block parsing) are definitely the first line of defense to keep code examples intact. but for long explanations, we evaluate local context shifts dynamically so split examples don't lose their target variables. how are you handling multi-language or mixed code/text docs on your end?