Map a portfolio company in 7 minutes.
Show your IC the 90-day plan.
Upload a few company documents. The Intelligence Graph extracts the operational ontology, maps workflows, identifies AI value-creation opportunities, scores AI readiness, and produces a board-ready 30 / 60 / 90-day roadmap — every claim evidence-cited, every recommendation governed by an explicit agent-deployment blueprint.
Built for
PE Operating Partners
Map a portfolio company in 7 minutes. Replace the $50K–$200K AI-readiness deck with a board-ready value-creation roadmap.
CFOs of PE-Backed Companies
See your own organization's ontology, workflow bottlenecks, and SOX-mapped controls before the operating partner asks.
Corporate Development
Post-merger integration prep — map the target's workflows and AI-deployment opportunities alongside the diligence pack.
Mid-Market CIOs
Defensible AI deployment roadmap with evidence-cited opportunities, scoped agent blueprints, and governance built in from day one.
Why this exists
Palantir proved the ontology-first architecture.
We automate the first 90 days of it.
Enterprise AI deployment is moving from experimentation to operations. OpenAI and Anthropic are raising multi-billion-dollar deployment vehicles. Palantir has shown that the winning architecture is ontology-first — but its model is still high-touch and expensive, with engagements measured in quarters, not days.
The Intelligence Graph automates the first mile: ingest a portfolio company's documents, build the operational ontology, identify risks and value-creation opportunities, and generate a governed agent deployment roadmap. Sold to PE operating partners as a portfolio-wide value-creation map. Days, not quarters.
Days, not quarters
6-stage pipeline runs in ~3–5 minutes per portfolio company. Total turnaround from upload to board-ready memo: under a day.
Replaces the readiness deck
Direct alternative to Big 4 "AI readiness assessment" engagements that cost $50K–$200K and take 6–12 weeks per portfolio company.
Built for the IC
Every claim cited, every confidence score visible, every blueprint governance-bounded. Output stands up to an LP-grade audit.
What this product actually does
Specific capabilities. Specific evidence. No marketing fluff.
Auto-Extracted Ontology
A proprietary six-stage agentic pipeline extracts the business-object ontology your operating-partner team would normally spend weeks building manually. Customer, Vendor, Contract, Invoice, Employee, Asset, System — every entity carries a source-document citation.
Evidence-Enforced (No Hallucination)
Every entity, risk, control, opportunity, blueprint, and initiative carries a verbatim quote from the source document plus a 0–1 confidence score. The pipeline refuses to invent findings it cannot cite — you can ship the output to your IC without re-reading the source data.
Regulatory Framework Mapping
Risks and controls are mapped to specific regulation references — EU AI Act Article 14 (human oversight), NIST AI RMF GOVERN 1.1, GDPR Article 32 (security of processing), SOX 404, ISO 42001 — not just framework names. AIGovHub-compatible output payload.
AI Readiness Score (0–100)
Composite score across data quality (35%), workflow documentation (25%), control maturity (25%), and cultural readiness (15%). The score gates the recommendation: PROCEED (≥65), CONDITIONAL (45–64), PAUSE (<45).
Governed Agent Blueprints
For every automation opportunity, a deployment blueprint: permission level (read-only / advisory / supervised / autonomous), allowed actions, blocked actions, data scope, human approver roles, audit retention horizon, recommended LLM. No autonomous money movement. Ever.
Top 10 Opportunities, Quantified
Each opportunity ranked by estimated hours saved and dollar savings per year, with risk classification and implementation timeline. The roadmap then sequences them across 30/60/90-day phases with EBITDA impact, cost, and payback months per initiative.
Multi-LLM Orchestration
Gemini 2.5 Pro for structured extraction, Claude Sonnet 4 for governance reasoning, DeepSeek V4 for cost-sensitive volume, Grok 4 for real-time intelligence. Model-agnostic — when the AI labs evolve, we swap providers without you noticing.
Graceful on Sparse Inputs
If your uploaded documents are thin (one-page profile + vendor list only), the pipeline returns a quality-flagged report with explicit "what to upload to deepen this analysis" guidance — never a confidently-wrong empty section.
The six-stage agentic pipeline
Ontology → Workflows → Risks & Controls → Opportunities → Blueprints → Roadmap.
Ontology Mining
Auto-extracts business objects (Customer, Vendor, Contract, Invoice, Employee, Asset, System) and their relationships from uploaded documents — every entity carries a source-document citation
Workflow Mapping
Identifies named processes (Procure-to-Pay, Order-to-Cash, Hire-to-Retire) with sequential steps, owner roles, system dependencies, approval thresholds, and bottleneck signals
Risk & Control Mapping
Maps risks (SOX, AI Governance, Vendor, Cyber, Privacy, Fraud) to existing controls — with regulation references (EU AI Act Art 14, NIST AI RMF, GDPR Art 32, ISO 42001) and maturity scoring
Automation Opportunity Scoring
Scores each workflow into one of four buckets (read-only / human-in-loop / supervised / autonomous), estimates hours saved and dollar savings per year, recommends a specific agent name
Agent Deployment Blueprints
Per-opportunity governance blueprint — permission level, data scope read/write, blocked actions, human approver roles, audit retention requirements, recommended LLM, rollout phase
30 / 60 / 90-Day Roadmap
Board-ready value-creation roadmap with phased initiatives, estimated EBITDA impact, payback months, dependencies, success metrics — plus an overall PROCEED / CONDITIONAL / PAUSE recommendation
What you take to your IC
Four deliverables per run. Every one cited and reproducible.
Value-Creation Memo (PDF)
Board-ready PDF: executive summary, AI Readiness Score (0–100 with breakdown), 30/60/90 roadmap, top 10 opportunities table, risk + control matrix, workflow map, ontology, agent blueprints, methodology with confidence scores. Archive-safe for distribution to LP advisory committees and portfolio CFOs.
Editable Word Memo
Ready-to-redline Word document. Drop it on a portfolio CFO's desk, let them mark up the assumptions, and ship the final version to the operating partner review. Preserves all evidence citations.
IC Presentation Deck (PowerPoint)
Board-ready PPTX — title + recommendation, exec summary, AI Readiness Score, top opportunities, 30/60/90 roadmap, risk matrix, agent governance posture. Drop directly into your quarterly portfolio review.
Structured Excel Export
Multi-tab workbook (opportunities, blueprints, risks, controls, workflows, roadmap, ontology) for analysts who need to integrate findings into portfolio dashboards or cross-company AI readiness benchmarks.
Platform capabilities
Eight intelligence capabilities wired into one orchestrated pipeline — so your operating-partner team stops stitching tools together and starts ranking value-creation priorities across the portfolio.
Every capability is modular and provider-agnostic. As the AI labs evolve, we swap components under the hood without changing your interface, your deliverables, or the evidence citations your IC team already trusts.
Multi-LLM orchestration
A proprietary router picks the right model per stage — extraction throughput, governance reasoning, board-quality prose — so we never overpay for the wrong tool. Drop-in replacement when a new model lands.
agentskills.io-compatible skill specs
Every stage is encoded as a portable skill specification — version-controlled, Hermes-runtime-compatible. Self-hosted enterprise customers can move the pipeline into their own infrastructure without forking our codebase.
Evidence-enforced extraction
Pydantic-validated output schemas with per-finding confidence scores and verbatim source quotes. Speculative findings (<0.3 confidence) are suppressed from headline numbers. No hallucinated citations reach the report.
Regulatory framework engine
Risks and controls map to AI Act / SOX / NIST AI RMF / GDPR / ISO 42001 specific articles. Output payload is structured for direct consumption by compliance-management platforms.
Six-stage composable pipeline
Ontology → Workflows → Risks/Controls → Opportunities → Blueprints → Roadmap. Each stage outputs structured data that feeds the next. Failures degrade gracefully — partial output beats no output.
Document-to-graph transformation
Unstructured policy/SOP/vendor-list text becomes structured entity-relationship data, queryable by downstream analytics. Forms the foundation for future Phase-2 interactive graph visualization.
Provenance-first design
Every entity, risk, control, opportunity, and recommendation traces back to a specific source document with page number and verbatim quote. An IC partner can audit any number in the report in 30 seconds.
Quality gates on sparse inputs
Document-density pre-check estimates analysis depth before running the pipeline. Sparse uploads trigger explicit guidance on what to add ("recommended: SOX risk register, top-3 SOP exports") instead of producing thin, confidently-wrong output.
Map your next portfolio company
Send us a few documents. In under 7 minutes you'll have a board-ready value-creation memo with evidence-cited opportunities, governed agent blueprints, and a 30 / 60 / 90-day roadmap.