knowledgelib.io — AI Knowledge Library

Structured, cited knowledge units built for AI agents. Pre-verified answers to high-friction questions with full source provenance, confidence scores, and freshness tracking. One knowledge unit replaces 5 web searches and 8,000 tokens of parsing.

Get Started — API Docs Browse All Units

Why

When AI agents need to answer complex questions, they run multiple web searches, fetch noisy pages, and spend thousands of tokens parsing and cross-referencing — often producing unreliable answers with no source trail. The harder the topic, the worse it gets: a simple product question may cost a few thousand tokens, but a compliance ruling, ERP integration spec, or system design question can burn 30,000–80,000 tokens across 10+ searches and still hallucinate. knowledgelib.io short-circuits this:

Without knowledgelib.ioWith knowledgelib.io
Token cost8,000 – 80,000 tokens~200 – 4,000 tokens
Sources citedUsually none100% — every claim linked to a source
ConfidenceUnknownScored 0.0 - 1.0 per published methodology
Compute cost$0.50 – $15.00 per questionFree
FreshnessUnknownVerified date + update schedule on every unit

How It Works

Each knowledge unit is a self-contained answer to one canonical question. Units use YAML frontmatter for machine-readable metadata and inline source citations for auditability.

For AI agents (via web search — zero setup)

Every knowledge unit has its own web page optimized for search engine indexing. When your agent searches the web, our pages rank for AI-style queries (factual, specific, including year and data type). The page body IS the markdown — no HTML parsing needed.

For AI agents (via API / MCP)

# MCP Server (Claude, Cursor, any MCP client)
npx knowledgelib-mcp

# Direct API
curl https://knowledgelib.io/api/v1/query?q=best+wireless+earbuds+under+150

# n8n community node
npm install n8n-nodes-knowledgelib

# LangChain retriever (Python)
pip install langchain-knowledgelib

See the full API documentation for endpoints, MCP setup, and integration guides.

Knowledge Unit Format

Each unit follows one of six templates (product_comparison, software_reference, fact, concept, rule, erp_integration). Here is a concept unit — the richest format, designed to prevent agents from misapplying frameworks:

---
id: finance/saas-metrics/burn-multiple/2026
canonical_question: "What is the Burn Multiple and how is it calculated?"
entity_type: concept
confidence: 0.91
last_verified: 2026-02-14
temporal_validity:
  status: stable
  change_sensitivity: low
constraints:
  - "Only meaningful for venture-backed companies actively burning cash"
  - "Denominator (net new ARR) can be negative — ratio inverts and loses meaning"
skip_this_unit_if:
  - condition: "User needs a profitability metric for a profitable company"
    use_instead: "finance/saas-metrics/efficiency-score/2026"
inputs_needed:
  - key: stage
    question: "What is the company's funding stage?"
    type: choice
    options: [Series A, Series B, Growth stage]
sources:
  - id: src1
    title: "Measuring the Burn"
    author: David Sacks
    url: https://sacks.substack.com/p/...
    type: primary_research
    reliability: authoritative
---

# Burn Multiple

## Definition
The Burn Multiple measures how much a company burns to generate
each incremental dollar of ARR: net burn / net new ARR. [src1]

## Key Properties
- **Formula**: Net Burn ÷ Net New ARR
- **Good**: < 1.5x (Series B+), **Acceptable**: 1.5–2.5x
- **Advantage over Rule of 40**: works pre-profitability [src2]

## Constraints
- Only valid when net new ARR > 0 [src1]
- Seasonal businesses need trailing-twelve-month smoothing

## Anti-Patterns
### Wrong: Using gross burn instead of net burn
### Correct: Always subtract revenue from total burn first [src1]

## Comparison with Similar Concepts
| Metric          | Key Difference          | When to Use         |
|-----------------|-------------------------|---------------------|
| Burn Multiple   | Efficiency per $ of ARR | Pre-profit startups |
| Rule of 40      | Growth + margin balance | Post-revenue SaaS   |
| Magic Number    | Sales efficiency only   | GTM benchmarking    |

Available Knowledge Units (1164 units across 14 categories)

Consumer Electronics (81 units)

Audio

TVs

Monitors

Phones

Tablets

Cameras & Projectors

Gaming

Storage

Power

Automotive

E-Readers

3D Printing

Transport

Computing (23 units)

Laptops

Desktops

Peripherals

Networking

Home (52 units)

Smart Home

Kitchen

Appliances, Furniture & Sleep

Tools

Office

Security

Bathroom

Fitness (12 units)

Software (263 units)

Debugging

System Design

Patterns & Algorithms

Security

Migrations

DevOps

Other

Business (583 units)

Frameworks

Pricing

Transformation

Market Entry

M&A

Investment

GTM

Fundraising

Operations

Governance

ERP Integration

ERP Selection

Build vs Buy

Retail Transformation

Sales Ops

Marketing Ops

People Ops

Product & Tech

Strategy

Go-to-Market

Startup

Growth

Industry Benchmarks

Startup Readiness

Startup Planning

Market Research

Customer Research

Lead Generation

Startup Legal

Other

Finance (85 units)

Valuation

SaaS Metrics

Financial Modeling

Macroeconomics

SaaS Benchmarks

Financial Ops

Industry Benchmarks

Other

Compliance (33 units)

Privacy

Financial

AI

Employment

Tax

Startup Legal

Personal Care (6 units)

Baby (7 units)

Outdoor (10 units)

Garden

Grilling

Camping

Hiking

Optics

Travel (4 units)

Luggage

Accessories

Bags

Pet (4 units)

Dogs

Cats

Energy (1 unit)

Discovery Channels

AI agents can find knowledgelib.io through multiple channels:

ChannelHowSetup
Web searchPages rank for AI-style queries. Body is raw markdown.Zero
MCP servernpx knowledgelib-mcpOne-time install
REST APIGET /api/v1/query?q=...Zero — no key needed
n8nnpm install n8n-nodes-knowledgelibOne-time install
LangChainpip install langchain-knowledgelibOne-time install
CatalogGET /catalog.json — full index of all unitsZero
AI manifestGET /.well-known/ai-knowledge.jsonZero

Trust & Verification

Every knowledge unit includes: