Pinecone launches Nexus to boost AI agent task success

Pinecone has launched Nexus, a knowledge engine for AI agents intended to address low task completion rates caused by retrieval-heavy workflows.

The launch also includes KnowQL, a query language for agents; a new Marketplace; a Builder tier priced at USD $20 a month; native full-text search in public preview; and new cloud regions in Germany and Singapore.

Pinecone is positioning Nexus as a move away from conventional retrieval systems that return documents for large language models to process at inference time. Instead, Nexus compiles source data into structured, task-specific artefacts before an agent makes a request, aiming to reduce latency, token use and ambiguity in outputs.

According to Pinecone, many AI agents remain stuck at 50% to 60% task completion because they repeatedly retrieve and re-read raw content while trying to resolve gaps or conflicts in context. The process can consume most of an agent's effort and still often requires human review before action is taken.

Nexus has two main parts: a context compiler and a composable retriever. The compiler ingests source material and task specifications, then creates specialised contexts for different agent roles. The retriever serves those curated outputs in the format required for a given task.

Pinecone said examples include sales, finance, marketing and chief executive workflows drawing on the same underlying systems but receiving different artefacts. Those systems can include data warehouses, Salesforce, Slack, Gong, Gmail, Jira and Google Drive.

Governance features built into the system include role-based access controls, versioning, traceability to source data and rules for handling personally identifiable information. The system also includes a dashboard for usage, spending and compliance monitoring.

Query layer

Alongside Nexus, Pinecone introduced KnowQL, a declarative query language intended to give agents a standard way to request information. The language is built around six primitives: intent, filter, provenance, output shape, confidence and budget.

Pinecone argues that agent developers still rely on custom integrations and retrieval logic because there is no common interface for knowledge access. It drew a parallel with SQL, which standardised how applications interacted with relational databases.

Harrison Chase, Chief Executive Officer of LangChain, commented on the work with Pinecone: "Building reliable, long-horizon agents is fundamentally a context engineering problem. Getting the right information to the agent in the right format is what separates demos from production. Pinecone Nexus solves that at the knowledge layer, and KnowQL is the standard interface the agentic ecosystem has been waiting for. That's why LangChain is collaborating with Pinecone to advance it. And as teams build agents with Nexus, LangSmith gives them the observability and evals to help them move through the agent improvement loop."

Several partners are backing the launch with integrations or joint development work, including Box, Unstructured, Teradata, LlamaIndex and ThoughtFocus. Use cases span enterprise content, data ingestion, observability and vertical applications.

Tamar Bercovici, VP of Engineering at Box, said: "For AI agents to have a tangible impact, they need a seamless and secure way to work within your business. This means giving agents the necessary context found in the content organizations rely on daily - from product roadmaps and research to marketing assets and countless other file types. By securely integrating the decades of enterprise content that Box manages with Pinecone Nexus' Knowledge Engine, we're giving AI agents the context they need to deliver more accurate and efficient results."

Brian Raymon, CEO of Unstructured, said: "Enterprise AI lives or dies on the quality of the data feeding it. The hardest part has always been transforming the messy, document-heavy reality of enterprise content into something agents can actually reason over - and 87% of the Fortune 1000 trust Unstructured to do exactly that. Connecting Unstructured's ingestion and preprocessing platform to Pinecone Nexus closes the gap between raw enterprise data and the trusted, task-specific knowledge agents need. Together, we're turning every organization's unstructured data into a knowledge asset that compounds with every agent interaction."

Sumeet Arora, Chief Product Officer at Teradata, said: "Enterprises win with agentic AI when their agents can reason over the full landscape of trusted business data. Teradata is partnering with Pinecone on KnowQL and Knowledge engine, giving the world's most demanding organizations a clear path from governed data to autonomous, accurate, agent-driven outcomes - backed by the scale and trust enterprises have built on Teradata."

Marketplace and pricing

Pinecone also launched Pinecone Marketplace with more than 90 knowledge applications across sales, insurance, real estate, legal, human resources and customer support. According to the company, the applications can be deployed and modified without assembling separate ingestion, embedding, retrieval and orchestration layers.

The new Builder tier is priced at USD $20 a month and is aimed at developers and small teams. Pinecone also introduced Dedicated Read Nodes for sustained, higher-volume retrieval workloads, and said Bring Your Own Cloud deployments would support customers with data residency or regulatory requirements.

As part of the wider platform update, Pinecone integrated native full-text search into its database, combining vector and keyword retrieval in one system. The company said the change reflects lessons from building Nexus and customer demand for agent-scale query handling.

Early access for Nexus and KnowQL is open to customers and partners in sectors including financial services, healthcare, legal and enterprise software. Pinecone said measured results from the new approach include task completion rates above 90%, time to completion up to 30 times faster and token spending reduced by as much as 90%.

Jerry Liu, CEO of LlamaIndex, said: "2026 is the year agents move from workflows to employees - and the bottleneck is no longer the model, it's getting agents grounded in the right knowledge. We're excited to bring LlamaParse's document ingestion and parsing into Pinecone Nexus and the Pinecone Marketplace. Teams can now go from messy documents with complex layouts, tables, handwriting, and images to trusted knowledge ready for agents to act on - whether they're shipping a Marketplace app or building on Nexus directly."