Patent search
1,800 technical spec documents · 2.3 GB
- Fewer tokens
- 97%
- Faster latency
- 77%
- Accuracy improvement
- +25%
Pinecone Nexus
Give agents knowledge
Nexus compiles data into a queryable knowledge layer. One declarative query returns the full answer.
Nexus makes agents more accurate, faster, and cheaper
Nexus + KnowQL
Time0.00s
Tool Calls0
Tokens0
$ agent --system nexus-knowql
Traditional Agents
Time0.00s
Tool Calls0
Tokens0
$ agent --system traditional-agentic-rag
How it works
01
Data in
Nexus connects to existing data sources via Connectors. No restructuring required.
Data sources
Docs & wikis
Databases
APIs & SaaS
Object storage
Code repos
Connectors
Pinecone Nexus
knowledge layer
02
Compiled once
A Manifest defines what to extract. Nexus builds structured knowledge from the source corpus — one time.
Raw corpus
spec_0001.pdf
spec_0002.pdf
spec_0003.pdf
transcript.json
claims.xlsx
+1,795 more
Manifest
v1claim_idstring
filed_datedate
assigneeentity[]
prior_artcitation[]
statusenum
03
One call per query
Agents query via KnowQL. One typed call replaces the entire retrieval loop.
KnowQL
query PatentKnowledge {
prior_art(
claim: $claim
)
}
Agentic RAG
embed(query)
vector_search(k=20)
llm_reason(chunks)
vector_search(k=20)
llm_reason(chunks)
vector_search(k=20)
llm_reason(chunks)
llm_assemble(context)
Tokens201K
Latency4 min
Steps8+
04
Grounded answers
Every response is typed, cited, and scoped to what the agent is permitted to see.
Answer · prior_art
Based on claims 1–4 of US10,234,891, prior artfound in EP2,891,203 (filed 2014-03-12) andJP2016-084521 directly anticipates element 2(b).Patentability risk: HIGH.
typedcitedaccess-scoped
The results hold across corpus types
We benchmarked Nexus against agentic RAG across three different industries. The pattern held regardless of corpus shape, query type, or domain.
0%Faster latencyvs. agentic RAG
0%Fewer tokens per queryvs. agentic RAG
0%Accuracy improvementvs. agentic RAG
M&A due diligence
90 documents · 10 folders · PDF, XLSX, markdown
- Fewer tokens
- 92%
- Faster latency
- 48%
- Accuracy improvement
- +14%
Revenue intelligence
217 call transcripts · structured JSON · 45 MB
- Fewer tokens
- 85%
- Faster latency
- 18%
- Accuracy improvement
- +94%
Nexus benchmarked against hybrid retrieval with agentic RAG loop. Accuracy scored by LLM judge on 0–1 scale against human-labeled expected answers.
KnowQL
Before SQL, every application built its own data access layer. Agents face the same structural moment. KnowQL gives them a shared vocabulary that provides structured, precise, and grounded knowledge in a single call. These are the six primitives that allow agents to compose across heterogeneous knowledge sources.
ask
The goal, output schema, and contexts to query.
where
Deterministic predicates and access-control enforcement.
ground
Field-level citations with confidence tiers.
shape
Typed fields returned exactly as the agent specified.
confidence
Grounded assertions separated from uncertain inferences.
budget
Depth tier and latency envelope, plus a token budget.
Give agents knowledge today
Pinecone Nexus is now in public preview