PivotalEnterprise knowledge platform powered by your choice of LLM. Not stateless RAG: documents are ingested, entities and concepts extracted, and a persistent wiki built that grows with your organisation. Knowledge stays current, searchable and auditable.
Pivotal handles ingestion, extraction, organisation and retrieval. Your team focuses on using knowledge, not maintaining it.
Not stateless RAG. Pivotal builds a persistent wiki from your documents. Nine typed pages - entity, concept, person, decision, summary, index, overview, category and section - accumulate and organise knowledge automatically. Curators can lock a placement so the LLM cannot override it.
On ingestion, your chosen LLM extracts entities, concepts and relationships automatically. The wiki builds itself as documents flow in, without manual curation overhead.
Define agents as ordered recipes of skills - read, search, synthesise, propose edits, route to people. They run on a schedule, on demand or when knowledge events fire, and every proposed change passes through a human review queue before it reaches the wiki.
Pivotal surfaces what is missing: topics referenced across many documents with no page, people mentioned often but never profiled, and under-represented knowledge areas. Turn any insight into a wiki stub with one click.
Every page carries a grounding score - the fraction of its citations that remain valid after reingest. A nightly LLM audit adds findings for contradictions and stale facts, so you can see at a glance which pages to trust and which need attention.
Entities and concepts connect in a browsable graph and topic map. Explore relationships visually, trace concepts across documents and understand how knowledge is distributed.
pgvector semantic search and full-text retrieval run concurrently across both wiki pages and document chunks. Reciprocal Rank Fusion merges the results and FlashRank reranks the set, finding answers buried across hundreds of documents with the precision of structured retrieval.
Ask questions in plain language, get streamed answers grounded in your own documents, with inline citations back to source. No hallucinated references; every answer is auditable.
Pivotal is purpose-built for teams where institutional memory is a competitive advantage.
Accumulate project learnings, client context and methodologies. Every engagement enriches the knowledge base.
Precedents, policies and regulatory knowledge linked and searchable. Never lose a decision or its rationale.
Research, analysis and investment theses organised by entity. Analysts find context without hunting through archives.
Technical expertise and project history preserved and retrievable. Onboarding new staff becomes measurably faster.
Pivotal is a knowledge application. It sits above the gateway and inference layers, using LLMs to build and query organisational memory.
Pivotal is an LLM-powered application. Alloy can sit in front as the gateway, managing which providers Pivotal uses, enforcing cost budgets and capturing audit logs of every inference call.
Kaizen is an AI coding agent for software teams. Pivotal is an organisational knowledge platform for any knowledge-intensive team. Both are LLM-powered applications; neither is a gateway or inference engine.
Pivotal focuses on building and querying organisational knowledge. AgentOS orchestrates autonomous agents across enterprise workflows. Pivotal can be a knowledge source for AgentOS agents.
Pivotal is an agentic enterprise knowledge platform. It ingests documents such as PDFs and DOCX files, uses your chosen LLM to extract entities, concepts and relationships, and builds a persistent wiki of typed pages that compounds over time. Unlike stateless RAG, every document that flows in adds to a shared knowledge structure your team can search, browse and audit.
Pivotal is designed for knowledge-intensive teams in consulting, legal, financial services and professional services where institutional memory is a competitive advantage. It suits organisations that struggle to find decisions, precedents or expertise buried in document archives, or that lose knowledge when team members move on.
Pivotal entered Early Access in Q2 2026, with a Preview release targeted for Q3 2026 and RTM in Q4 2026. Join the waitlist to access the current early access build.
Standard RAG is stateless: every query searches documents cold with no accumulated understanding. Pivotal builds a persistent wiki from your documents, extracting nine typed page types including entities, concepts, decisions and summaries. Knowledge compounds with every ingestion, a grounding score tracks citation validity, and a nightly LLM audit flags contradictions and stale facts so you know which pages to trust.
Pivotal features LLM-powered entity and concept extraction on ingestion, an agentic workflow engine with skill-based recipes and a human review queue, hybrid search using pgvector and full-text retrieval fused with Reciprocal Rank Fusion and FlashRank reranking, a knowledge graph and topic map for visual exploration, multi-space support for organisation and project spaces, and chat with inline citations grounded in your own documents.
Join the Pivotal early access waitlist to register for the current access build. Join Early Access.
Pivotal is in early access from Q2 2026. Join the waitlist to be among the first teams to deploy a knowledge base that compounds.
