Enterprise Knowledge Management

Notaknowledgebase.A Cognitive OS.

A three-layer architecture: a data layer that keeps the material clean, a search engine that finds it all and answers with evidence, and a feedback loop that makes it more accurate the more it's used.

The big picture

Three layers, one loop

These layers are not one-way. Feedback flows back into the data layer — a closed loop, and the biggest difference between this and a system you build once and never touch again.

Layer 1
Data Infrastructure

Defines the knowledge and storage structure so retrieval can be accurate — where data lives, how it's classified, processed, and stored.

Layer 2
Search Engine

The brain. Its core loop is Collect → Act → Verify: it searches on its own, and it verifies its own evidence before answering.

Layer 3
Feedback & Governance

Collects real usage signals — ratings, conflicts, evaluations — and feeds them back into the data layer.

The closed loop: clean data → precise retrieval → verified answer → real-usage feedback → cleaner data
Layer 1 · Data Infrastructure

Good retrieval starts with good data

Tens of terabytes across legacy Notes archives, SharePoint, file servers, CMS, and engineering & records systems. The point is not how big it is — the point is how the engine can pinpoint things inside it.

KB Classification

Dump everything into one pool and every search digs through the whole warehouse. So we split it: top-level KBs by business area, sub-KBs by topic — and every KB states a name + description, written for the engine. A question comes in, the engine reads the descriptions, picks the right KB, and searches only there. Like a library: read the section signs first, then walk to the right shelf.

Standardized Pipeline

Every piece of data goes through the same line: define the type → tag it → clean it → decide how to store it. Tagging is the key step— department, project, year, status, access rights. With good tags the engine narrows the search automatically, and access control runs on the same tags: no permission means the file doesn’t even show up.

Fine-grained Extraction

We don’t dump raw full text in. Each file type gets the structure that fits it — so a question about one clause returns that clause, not everything that sounds similar. Cleaning handles duplicates, format issues, version relationships, and removes outdated content.

Each data type gets the structure that fits it
Emails

Structured JSON — sender, recipient, subject, body, attachments as separate fields. Find a mail by person or topic, and locate its attachments.

Legal documents

Clause-level JSON, layered document → chapter → clause. Ask about eligibility criteria and get clauses from that chapter only — nothing similar-sounding leaks in.

Tables

Relational records — queryable rows and columns, not flattened text.

Drawings

Vector JSON — geometry and annotations preserved for spatial questions.

Documents

Markdown with position info — every passage knows exactly where it came from.

Atomic Retrieval Capabilities

Six tools, one toolbox

On top of the clean data sit six standard retrieval capabilities. The engine above picks whatever tool the question needs.

Knowledge Base Search

Vector + BM25 + knowledge graph, with metadata and permission filtering built in.

Database Query

Structured questions answered from live business tables.

Full-Text Search

Character-exact lookup across the whole corpus.

API Calls

Pull answers from your existing systems in real time.

MCP Calls

Standard tool protocol — new capabilities plug in without rework.

Trusted Web Fetch

Whitelisted external sources only — the open web on your terms.

Technical Architecture

Layer 2: the engine — Collect, Act, Verify

The first stages find more — that's recall. The gate and the filtering choose better — that's precision. Every answer is traceable, and the system would rather say 'not found' than invent one.

Stage 01

Understand & Rewrite

The query goes through a glossary first — casual phrasing, jargon, abbreviations rewritten into something the engine can use. If the question is unclear, it asks the user. It doesn't guess.

Under the Hood

Intent detection + term normalization; clarifying questions instead of low-confidence guesses.

Stage 02

Plan the Retrieval

The engine picks the best target KBs by intent — reading the names and descriptions the data layer defined. It never broadcasts the question to every KB.

Under the Hood

Like a library: read the section signs first, walk to the right shelf. KB descriptions are written for the engine.

Stage 03

Isolate at the Entrance

Before the KB is even entered, ACL and metadata lock the scope down. Content the user can't see never enters the candidate pool — security is isolation at the entrance, not filtering the answer at the end.

Under the Hood

Permission tags + metadata (year, region, status) narrow the search space before retrieval starts.

Stage 04

Hybrid Recall

Three paths in parallel: BM25 for exact matches, vectors for meaning, the graph for relationships. Results merge and dedupe by chunk ID — say 30 + 50 with 10 overlapping = 70 unique candidates.

Under the Hood

Each path owns one job the other two can't do; each blind spot is covered by another. Recall: find everything, miss nothing.

Stage 05

Rerank + Dynamic Top-K

Candidates are rescored against the question one by one; then the scores decide the cutoff. 0.92, 0.88, 0.85, then a cliff to 0.41? The line is drawn at the cliff — above it goes to the LLM, below is thrown away.

Under the Hood

K is not hard-coded: a simple question keeps 3, a complex one keeps 8. Irrelevant context is noise that breeds hallucination.

Stage 06

Sufficiency Gate → Generate

Evidence is checked before any answer: conflicts resolved toward the fresher, more authoritative version; expiry, permission, completeness verified. Enough evidence → generate with citations. Not enough → re-route, re-retrieve, or pull the full text and go again.

Under the Hood

Max-rounds and token caps prevent endless loops. Dead end → “I can't find it.” It never makes things up.

Hybrid Recall

Three paths, zero gaps

Each path is the best at one thing — and blind at another. Use any single one and you lose a whole class of questions. Run all three in parallel — no gaps.

Retrieval pathIts valueIts blind spotCovered by
BM25 keywordNever misses an exact match — invoice numbers, clause numbers, legal terms.Change the wording and it's lost: ask “when does the customer pay”, the document says “payment terms” — no match.Vector search catches the rephrasing.
Vector semanticUnderstands meaning, not words — casual, formal, or another language; same idea, found.Unreliable on precise items: INV-12345 and INV-12346 look almost identical to a vector model.BM25 nails exact IDs; the graph separates confusable entities.
Knowledge graphFollows relationships — “the partner's IT staff” lives in connections, not in any single passage. Multi-hop: partner → IT staff → operations → outsourced.Only does relationships — regular content questions aren't its job.BM25 and vector handle the content.
The key checkpoint

The Sufficiency Gate

Evidence is checked before we answer — not after.

Conflict check

Two documents disagree? Trust the fresher, more authoritative version — and log the conflict for governance.

Availability check

Is it expired? Does this user have permission? Is it complete? Fails here never reach the answer.

Source retention

Every piece of evidence keeps its origin and link — ready for citation in the final answer.

Two exits. Evidence is good — generate the answer, with citations. Not enough — loop: change direction, switch tools, pull the full text, try again (with max-rounds and token caps, so it never loops forever). And if it’s truly a dead end — the system says “I can’t find it.” It never makes things up.

See it answer

Cited answers, live graph

Ask in plain language — get the answer, the sources it came from, and the knowledge graph behind it.

LIVE DEMO
KB
GPTBots Knowledge AI
Graph-RAG · 2 hops · 4.2s indexed
Retrieval Trace
Vector Match12
Graph Traversal2 hops
Confidence94%
AUDIT-TRACEABLE · 3 sources cited
Sources Retrieved
Policy DocsRefund-Policy-v3.2
SOP LibraryFinance-SOP-2026
Case HistoryCase #A-1182
AccessSales Ops + Finance
Last Sync12 minutes ago
AI Workflow Layer

Search is a capability. Many scenarios are workflows

Not everything is Q&A. The AI Workflow Layer builds flows from your own SOPs — calling the LLM, intent detection, and knowledge retrieval at exactly the right steps — to deliver the process the business needs, not just an answer box.

Proven at national scale

In one national-scale deployment we mapped 33 user stories — every single one:

8
search-type stories → handled by the search engine
25
business scenarios → AI workflows + custom builds
Layer 3 · Feedback & Governance

The more it’s used, the more accurate it gets

Three feedback streams — two active, one passive — flow back into data quality, labels, knowledge structure, and retrieval strategy. This system is not static.

User active feedback

Likes and ratings flow straight back into data quality and labels. What gets praised and what gets thumbs-down becomes governance input.

Passive conflict detection

During retrieval the system catches documents that contradict each other — an old and a new policy version — flags the fragments, and hands them to a human to confirm which to trust. Users do nothing.

Scaled A/B evaluation

Real user questions re-run regularly against KB v1.0 vs the improved v2.0. Results prove whether each change actually helped — and keep tuning the retrieval strategy.

Bottom to top: the data layer keeps the material clean · the engine finds it all, picks the best, and answers with evidence · the feedback loop keeps it improving
Platform ROI

From process-driven to cognition-driven

40TB
Knowledge scale proven

One deployment: 40TB across legacy archives, SharePoint, file servers, CMS, and records systems.

-70%
Time finding & drafting

Specialists stop being retrievers and drafters — they become decision makers.

100%
Answers with citations

Every response is grounded in retrievable evidence with source links.

0
Invented answers

The Sufficiency Gate would rather say 'not found' than guess — grounding is enforced.

Bring your messiest knowledge problem

Policies in five systems, answers in none? We'll show trusted, cited answers on your own documents.