ashashutosh

Another milestone in our @pinecone Everywhere mission: cut the time from data to outcome, lower the TCO.

Today we added @Google Gemini Embedding 2 to Pinecone integrated embedding. A few days ago it was OpenAI. Building and running an embedding pipeline is a necessary part of your AI stack, so we manage all of it for you. You pick the model and build the app.

This one is special for me, not just as a Xoogler. @Google Cloud is a key hyperscaler partner for us, with a large share of our biggest enterprise customers already running on their AI platforms.

Using it: set gemini-embedding-2 on a text field and send plain text. Pinecone embeds every write and every query and runs the search in the same call. We serve the model from Google Vertex AI and retry failed calls, fully transparent to your code.

No retry or failover code. No Google Cloud account or API key. No second bill. The tokens show up on your Pinecone invoice at Google’s price.

100+ languages. Pick the task the model tunes for, from search to Q&A, classification, or code retrieval, and Pinecone adds the right instruction. Keep your dimensions, so it drops into indexes you already run.

Pinecone Everywhere series so far: pg_pinecone put vector search in Postgres. BYOC put the platform in your own secure cloud. Integrated embedding takes a core stage of the stack off your plate.

One mission behind all of it: lower time-to-outcome and lower TCO for building and running agentic apps.

#Pinecone #GoogleCloud #Gemini #EnterpriseAI #PineconeEverywhere

A theme runs through everything we build at @pinecone : efficiently reduce time-to-outcome.

The Pinecone Everywhere series has been about where your knowledge can live. In the Postgres you already run. In your own cloud. Today is the other half of the problem: how much work it takes to get your data into AI search at all.

Running semantic search on an OpenAI model used to mean owning the account and key, the rate limits, the failover, and a second bill every month. This month, next month, forever.

Now Pinecone integrated embedding runs OpenAI’s text-embedding-3 models for you. Set the model on a field, send plain text. We embed every write and every query, fail over across Azure regions on our side, and put the tokens on your Pinecone invoice at OpenAI’s list price. No OpenAI account, no key, no separate bill.

pg_pinecone put search in your Postgres. Bring Your Own Cloud put it in your cloud. Integrated embedding takes a whole pipeline off your plate. Same goal every time: less for your team to own, faster time-to-outcome, lower TCO.

#Pinecone #EnterpriseAI #OpenAI #EnterpriseAIKnowledgePlatform

Last week: pg_pinecone, our first surface, the search engine we run at scale, inside the Postgres you already have. Yesterday: where it fits and how it scales.

Today, the second surface in @Pinecone Everywhere: the cloud account your knowledge is required to stay in.

AI’s value scales with how much of your own knowledge it can use: how the work gets done, the history, the policies, years of expertise. For regulated workloads, that knowledge is a top asset, and it has to stay inside the firewall.

That left two bad options. Run the knowledge infrastructure yourself. Or hand the data to a managed service and give up control.

Pinecone BYOC, Bring Your Own Cloud, is a better way, built with our largest security-sensitive customers. The data plane, where your knowledge lives, runs in your own cloud. We manage the control plane to keep it running.

No SSH, VPN, inbound access, or standing cross-account role. Maintenance is an outbound call from our control plane, run locally. No persistent path into your environment. Works with SSO, RBAC, SCIM, SAML, audit logging, encryption, and private networking.

Same Pinecone APIs and SDKs across managed and BYOC. Critical workloads stay in your environment, others run managed, no separate code path. Move between the two as the business changes.

More on how the zero-access model works soon.

pinecone.io/blog/byoc-gene…

#Pinecone #BYOC #EnterpriseAIKnowledgePlatform

A year ago, the hardest question in enterprise AI was which model to use. Thanks to incredible work by model researchers, new capabilities arrived almost daily.

That question is largely settled. Models are capable, plentiful and increasingly interchangeable.

The hard question now: can the model reach your knowledge, securely? Your policies, your history, your records. The context that makes an answer right for your business and wrong for everyone else’s.

That is where most AI projects stall. Not on reasoning. On knowledge.

A year ago, we saw reasoning models and agents consuming the majority of Pinecone’s database APIs, with more AI going into production everywhere, at every size, on every platform.

So we spent this year building around one idea: knowledge has to be available wherever AI work happens.

- Any user: developer, platform team, or autonomous agent.
- Any surface: our managed service, your cloud, on-prem, edge, or inside the Postgres you already run.
- Any scale: a thousand vectors to billions in one index, at the latency, recall and cost production AI demands.
- Any model, any source: no moving your data, no model lock-in.

We call this the Enterprise AI Knowledge Platform, delivered through Pinecone Everywhere.

Over the next few weeks I will break down each piece with the numbers: Postgres, your own cloud, dedicated scale, integrated embedding and reranking, and Nexus, the knowledge engine for agents.

Tomorrow: the database most of you already run.

I would rather show the work than talk about it. Tell me what we have right and, more importantly, what we could do better.

#EnterpriseAI #AgenticAI #PineconeEverywhere

PostgreSQL is the industry-standard default for building the vast majority of modern applications, and also the default choice for coding agents. For experimentation and small agentic apps, we get a real database in seconds and we just build.

The trouble starts when those apps scale. pgvector, the common vector store for Postgres, is great for a prototype or a starter app, but hits ceilings on scale, performance, capability, and cost. Teams start looking elsewhere, and sometimes move off Postgres entirely.

We are big fans and users of Postgres, so we built pg_pinecone, a native Postgres extension, open sourced under Apache 2.0, integrating @pinecone search engine into Postgres.

The experience stays pure Postgres, technically and commercially: free and open source, with much better performance and scale at significantly lower cost. You keep the database you have and love, and get search that scales all the way into enterprise scale production.

Oh, and we added the one thing our trial users kept asking for: filtered search. If you are multi-tenant, the usual Postgres vector store gets less accurate the moment you add that filter, so a search narrowed to a single account can silently drop results it should have returned. pg_pinecone applies the filter during the search, so a per-customer query comes back complete.

When a table finally outgrows a single server, seamlessly move its search to Pinecone’s serverless database with one SQL command, with no changes to your queries.

Coding agents put a whole generation of apps on Postgres. Today, we let those apps grow up without having to leave it.