Pinecone is the trusted AI knowledge company. Pinecone's mission is to make AI knowledgeable.
United States
- Tweets 2,561
- Following 147
- Followers 27,760
- Likes 2,962
Nexus is GA. Enterprise knowledge, compiled once, query-ready for agents. On Sierra's τ-Knowledge: GPT-5.2 +12% accuracy, -80% cost. GPT-5.5 same accuracy, -77% cost.
Read the full announcement: pinecone.io/blog/pinecone-ne…
7
7
30
4,423
Toyota Motor North America has decades of manufacturing knowledge, but using it with AI came with one requirement: it had to stay inside Toyota’s environment
Pinecone BYOC runs there, while we manage upgrades and scaling without inbound access.
BYOC is now GA 🔗 pinecone.io/blog/byoc-genera…
7
1
14
925
Pinecone introduces VQ-bench, an open-source framework for testing and comparing vector quantization methods. See how it makes quantizers easier to build, benchmark, and evaluate across metrics like recall, reconstruction error, and performance.
Read the full breakdown on the Pinecone blog:
lnkd.in/dwMvF4uP
2
1
8
1,335
Use @typesafeai's Jev model with Pinecone to rerank results with natural language criteria!
Usually with rerankers, it's hard to cleanly specify what should and shouldn't be returned in results.
Jev resolves this by refactoring the problem into evaluating against distinct binary criteria, which pairs great with Pinecone retrieval!
In this demo, we compare using Jev and Claude to rerank 200 returned candidates from Pinecone. Jev returns a reranked list in about a second — 830 to 1,300 ms across eight test queries.
Claude Opus 5, doing the same job in one long-context call, takes 4.2 to 6.8 seconds.
That's roughly 5x faster, and about 43x cheaper: $0.004 per query against $0.18.
7
5
26
2,504
A new release of our Agent Skills and Docs eval library, Cultivar is available!
Now, use @typesafeai's Jev model to evaluate how well your agents use docs and Skill against specified tasks in Modal sandboxes.
This works great, because Jev allows for binary yes/no decisions calibrated with respect to criteria, exactly the format Cultivar works best with!
Preliminary results benchmarking grading agent runs with Jev over Claude produces a 19x speedup and is 38x cheaper.
4
2
6
1,018
However, using Claude allows for detailed reasoning and justification for each run. Use Claude for complex evaluations and Jev for quick ones!
Install and use cultivar here: github.com/pinecone-io/culti…
359
Dozens of vector quantization papers come out every year. When we surveyed them, we noticed that most quantizers are not new algorithms. They are the same handful of primitive operations, strung together in a different order.
So we built VQ-bench: an open-source library of those primitives, composable into pipelines. E-RaBitQ, one of the strongest methods we tested, is four primitives in a list. Swap one and you have a new quantizer, evaluated exactly the way every other method is evaluated.
We used it to benchmark 14 popular quantizers on recall, reconstruction error, and encode time. Two takeaways so far: PQ and OPQ have the lowest reconstruction error, and EDEN matches E-RaBitQ on recall while encoding much faster.
This is a first iteration. We want your feedback, corrections, and contributions, and we will keep adding quantizers over time.
Website: vq-bench.com/?utm_campaign=e…
Blog: pinecone.io/blog/vq-bench/?u…
Repo: github.com/pinecone-io/vq-be…
Paper (VecDB@VLDB 2026): arxiv.org/abs/2608.11240
5
2
13
990
A database knows where your data is. A vector database understands what your data means.
It finds information based on similarity and meaning, not just exact keywords, powering semantic search, RAG, recommendations, and agent memory.
See how it works: pinecone.io/learn/vector-dat…
7
969
Most of the attention on full-text search goes to BM25. Query a part number, get that part number back first.
A query like "why does PROD-001 overheat under load" has a literal string and a question in it though. That's where text-match filtering comes in. It narrows results to the documents that contain the literal string, and semantic search ranks what's left, all from one index.
Full-text search is GA in Pinecone Database, text-match filtering included.
🔗 pinecone.io/blog/full-text-s…
2
13
948
The LA Agentic AI Meetup in on the Tech Week calendar. Join us Tuesday, October 13, 5 to 7pm at Gulp in Playa Vista, for a few live demos, drinks, and real people comparing notes on what they're building and where things are going.
RSVP: partiful.com/e/bh5rBRPMPnOP9…
@Techweek_
1
2
8
923
🚀 Full-text search is now GA in Pinecone Database.
Semantic search doesn't find literal strings well. Query string literal like a part number and all the similar part numbers score about as high as the one you actually wanted, and nothing in the results tells them apart.
Full-text search solves this. BM25 keyword ranking runs in the same index as your vectors. So, exact matches are returned and ranked at the top of the list.
Link in the replies.
6
14
1,209
Pinecone retweeted
Jörg Schad is the VP of Engineering at @pinecone. He joins @kbal11 to discuss moving beyond RAG with precomputed context, context artifacts, semantic layers, and AI retrieval.
@joerg_schad
softwareengineeringdaily.com…
2
9
939
Next Tuesday, September 8, we're hosting the September LA Agentic AI Meetup, 5 to 7pm at Gulp in Playa Vista.
Come hang out with builders, engineers, founders, and the AI-curious for networking, live demos, drinks, and real conversations about what people are building in AI.
RSVP: luma.com/wzf6hq1d
2
6
1,008
Full write-up here 🔗 pinecone.io/blog/support-age…
1
218
We just added a "Start demo" button to our @AWS Marketplace listing.
Click it, get a pre-provisioned AWS environment, and start building on Pinecone immediately.
Zero setup. Zero installation. No provisioning required.
Try it out and let us know what you think: aws.amazon.com/marketplace/p…
1
10
1,035
API design has always had one reader in mind: a human developer. However agents are the new reader that looks at things a bit different. Here are some of the tenets that we now hold our API to:
1. Errors are guidance
2. Budget the reader's context
3. Self-description beats documentation
4. Safe at machine tempo
5. Access without a human in the loop
6. The agent surface is a product, not a mirror
Read more: pinecone.io/blog/designing-a…
2
1
13
1,083