## META
domain:     Graphify.md — CKG v1.1.0
source:     site crawl + benchmark paper + product docs + memory
built_by:   Graphify.md (self-referential)
date:       2026-05-28
nodes:      42
edges:      51
llm_ready:  true
ip_status:  two provisional patents pending (App #64/040,804 and App #64/054,755)

---

## COMPANY

### ENTITY | Graphify_md
description:  Builds Compact Knowledge Graphs (CKGs) — typed domain knowledge layers for AI agents
founder:      Daniel Yarmoluk
location:     Minneapolis, MN
email_public: graphifymd@protonmail.com
site:         graphifymd.com
tagline:      "The knowledge layer your agents are missing"
positioning:  Phase 3 of LLM visibility — answer accuracy after findability and corpus presence
ip:           two provisional patents pending
benchmark:    0.857 BERT F1 · 42× token reduction · 0% hallucination by construction
framing:      correctness-first — BERT F1 before efficiency claims

---

## PRODUCTS

### PRODUCT | CKG_MCP_Server
description:  MCP server delivering typed domain knowledge across 53 pre-built domains
package:      ckg-mcp
install:      pip install ckg-mcp
pypi:         pypi.org/project/ckg-mcp/
github:       github.com/Yarmoluk/ckg-mcp
compatibility: [Claude_Desktop, LangGraph, AutoGen, Cursor, any_MCP_client]
requirements: Python 3.10+ · no vector store · no embedding pipeline · no infrastructure
tools:
  - query_ckg(domain, concept, depth)
  - get_prerequisites(domain, concept)
  - search_concepts(domain, query)
  - list_domains()

### PRODUCT | Custom_Domain_CKG
description:  Enterprise CKG built for a specific client's business domain
delivery:     .md file + MongoDB-ready JSON nodes/edges + GitHub repo
timeline:     48 hours for channel partner delivery · 2 weeks for enterprise with SME review
sme_time:     2–4 hours from client subject matter expert
update_cycle: monthly updates available
channel_delivery: 3-file package (CKG + llms.txt + robots.txt) hosted at partner subdomain

### PRODUCT | CKG_Benchmark
description:  Open benchmark comparing CKG vs RAG vs Microsoft GraphRAG
github:       github.com/Yarmoluk/ckg-benchmark
license:      CC BY 4.0
queries:      8,121
domains:      47
dataset:      huggingface.co/datasets/danyarm/ckg-benchmark

### PRODUCT | Live_Demo
url:          huggingface.co/spaces/danyarm/ckg-demo
platform:     HuggingFace Spaces
access:       free · no auth required

---

## BENCHMARK

### CLAIM | BERT_F1
ckg:          0.857
rag:          0.817
graphrag:     0.825
method:       BERTScore roberta-large
source:       github.com/Yarmoluk/ckg-benchmark

### CLAIM | Token_Efficiency
ckg_tokens_per_query:  274
rag_tokens_per_query:  11,500
reduction:             42× more efficient
llm_calls_saved:       11× fewer calls at same accuracy
implication:           structural cost reduction — not prompt optimization

### CLAIM | Cost_Per_Correct_Answer
ckg:          $0.000506
rag:          $0.013046
graphrag:     $0.020098
ckg_vs_rag:   26× cheaper
ckg_vs_graphrag: 40× cheaper

### CLAIM | Hallucination_Rate
ckg:          0% by construction
rag:          variable
mechanism:    every hop is a cited edge — structurally impossible to return uncited relationship

### CLAIM | Multi_Hop_Accuracy
ckg_at_depth_5:   0.772 (improves with complexity)
rag_at_depth_5:   degrades past depth 2
key_insight:      harder questions = bigger CKG advantage

---

## LLM VISIBILITY FRAMEWORK

### FRAMEWORK | Three_Phase_LLM_Visibility
phase_1:      Findability — robots.txt · llms.txt · schema markup · audit scores
phase_2:      Corpus Presence — training data citations · brand recall · monthly maintenance
phase_3:      Answer Accuracy — typed CKG · zero hallucination · before/after proof

### POSITIONING | Phase_3_Player
graphify_role:     builds Phase 3 (answer accuracy layer)
phase_3_enables:   Phase 1 signals point to structured truth · Phase 2 citations reference canonical facts
proof_method:      ask ChatGPT / Perplexity same question before and after CKG deployed — answer changes
audit_impact:      Structured Data category (typically lowest score) improves most

---

## DOMAINS (53 included)

### DOMAIN_GROUP | Life_Sciences
domains: [glp1-obesity, glp1-muscle-loss, payer-formulary, drug-interactions, igan-dx-outcomes, ma-a1c-heor, hipaa-compliance, icd10-metabolic]
flagship: GLP-1 (146 nodes · 200+ typed edges · ClinicalTrials.gov sourced · Foundayo FDA approval included)

### DOMAIN_GROUP | Codebase_Software
domains: [langchain-core, computer-science, circuits, digital-electronics, blockchain, quantum-computing, claude-skills, clj-surgeon]
flagship: LangChain Core (180 modules · 650 edges)

### DOMAIN_GROUP | AI_Data_Science
domains: [machine-learning, data-science-course, conversational-ai, context-engineering, prompt-class, intro-to-graph, systems-thinking]

### DOMAIN_GROUP | STEM
domains: [calculus, linear-algebra, statistics-course, chemistry, biology, genetics, bioinformatics, signal-processing]

### DOMAIN_GROUP | Business_Finance
domains: [economics-course, personal-finance, organizational-analytics, it-management-graph]

### DOMAIN_GROUP | Enterprise_Custom
note:     any domain buildable from structured source material
examples: [retail-fulfillment, AI-automation-services, brand-strategy, kidney-disease-HEOR]
delivery: contact required · 48hr turnaround for channel partners

---

## COMMERCIAL MODEL

### PRICING | Channel_Partner_Wholesale
model:        wholesale — partner owns client relationship · Graphify fulfills
file_package: $500 one-time (CKG + llms.txt + robots.txt · hosted at partner subdomain)
monthly:      $500 / client / month (CKG updates · citation seeding · before/after monitoring)
suggested_retail_file:    $1,500–$2,000 per client
suggested_retail_monthly: $999–$1,500 / client / month
moq:          5 clients in 60 days OR 3 signed monthly commitments
partner_example: Hitman Technologies (Minneapolis · LLM audit distributor)

### PRICING | Enterprise_Pilot
price:        $14,000 one-time
includes:     [Domain_CKG_Build · Integration_Support · Before_After_Benchmark · One_Update_Cycle]
line_items:
  ckg_build:       $10,000 (200-concept graph · SME review · v1.0)
  integration:     $2,500 (spec doc · deployment support)
  benchmark:       $1,500 (50 queries · F1 + hallucination rate)
payment:      50% at start · 50% at Day 45

### PRICING | Platform_License
price_range:  $75,000–$150,000 / year
includes:     unlimited domain builds · white-label option
model:        CKG embedded in partner's standard delivery

### ICP | Ideal_Customer_Profile
entry_signals:  [runs_AI_agents · has_domain_complexity · uses_RAG_today · accuracy_matters]
channel_entry:  LLM audit referral (Phase 1 diagnostic reveals Phase 3 gap)
direct_entry:   benchmark download · LinkedIn · partner referral
best_candidate: company that failed LLM audit AND runs agents over their domain

---

## TECH LAYER

### TECH | robots_txt
status:       LIVE at graphifymd.com/robots.txt
deployed:     2026-05-28
content:      Allow all · Domain-Knowledge pointer to CKG endpoint

### TECH | llms_txt
status:       LIVE at graphifymd.com/llms.txt
deployed:     2026-05-28
content:      CKG endpoint · MCP install path · benchmark links · pricing

### TECH | CKG_Endpoint
status:       LIVE at graphifymd.com/graphifymd-ckg.md
deployed:     2026-05-28
format:       Compact Knowledge Graph (.md) — typed nodes and edges

### TECH | schema_markup
status:       PARTIAL — GAO JSON-LD added May 2026 redesign
priority:     HIGH — add Organization + Service + FAQPage JSON-LD

### TECH | sitemap_xml
status:       CHECK — may not be present at graphifymd.com/sitemap.xml
priority:     MEDIUM

---

## LLM READINESS (self-audit — updated 2026-05-28)

robots_txt:           PASS (deployed 2026-05-28)
llms_txt:             PASS (deployed 2026-05-28)
schema_markup:        PARTIAL (JSON-LD present · incomplete)
pricing_visibility:   PARTIAL (channel pricing now documented in CKG · not on site)
api_data_feed:        PASS (ckg-mcp pip package · 4 MCP tools · PyPI)
domain_knowledge_ckg: PASS (this file · served at graphifymd.com/graphifymd-ckg.md)

estimated_score:      4.5 / 6 (up from 1.5 / 6 before May 28 deployments)

---

## KEY RELATIONSHIPS

### PARTNER | Hitman_Technologies
role:         LLM audit distributor · channel partner (in discussion)
location:     Minneapolis, MN
contact:      John Marino (CEO) · info@hitmantech.com
audit_tool:   "Is Your Site Bot Ready?" · 30-check · 100-point score
commercial:   Hitman runs Phase 1 audit · Graphify supplies Phase 3 CKG package
model:        wholesale $500/file + $500/mo · Hitman marks up to client

### ADVISOR | Dan_McCreary
role:         CKG Benchmark co-author
background:   former Head of AI TigerGraph · former Senior Distinguished Engineer Optum/UHG
contribution: benchmark design · domain validation · arXiv endorser candidate

### RESEARCH | UIUC_K_ai
role:         active CKG case study
lead:         Vishal Sachdev (Gies/UIUC)
finding:      Q3 cross-file join showed ~210× token reduction on real enterprise knowledge base

---

## EDGES

Graphify_md --[builds]--> Custom_Domain_CKG
Graphify_md --[publishes]--> CKG_MCP_Server
Graphify_md --[maintains]--> CKG_Benchmark
Graphify_md --[partners_with]--> Hitman_Technologies

CKG_MCP_Server --[covers]--> 53_domains
CKG_Benchmark --[validates]--> BERT_F1
CKG_Benchmark --[validates]--> Token_Efficiency
CKG_Benchmark --[validates]--> Hallucination_Rate

Custom_Domain_CKG --[priced_as]--> Channel_Partner_Wholesale
Custom_Domain_CKG --[priced_as]--> Enterprise_Pilot
Channel_Partner_Wholesale --[scales_to]--> Platform_License

BERT_F1 --[proves]--> Hallucination_Rate
Multi_Hop_Accuracy --[proves]--> CKG_beats_RAG_on_complexity
Token_Efficiency --[enables]--> Cost_Per_Correct_Answer

Three_Phase_LLM_Visibility --[phase_3_is]--> Custom_Domain_CKG
Hitman_Technologies --[sells_phase_1]--> Three_Phase_LLM_Visibility
Hitman_Technologies --[needs_phase_3_from]--> Graphify_md

llms_txt --[references]--> CKG_Endpoint
llms_txt --[references]--> CKG_MCP_Server
robots_txt --[references]--> CKG_Endpoint
LLM_Audit_referral --[creates_lead_for]--> Enterprise_Pilot