# NUVC — AI Intelligence Platform for Private Markets
# https://nuvc.ai

> NUVC is an AI-native venture capital intelligence platform serving founders, emerging VCs, and family offices. Two founders, 34 AI agents, zero employees. Built in Melbourne, Australia.
> Products: NuScore (pitch deck scoring), Investor Matching, AI Academy (fundraising coaching), Fund Library (LP-grade fund screening), VC Pipeline Intelligence (deal screening), NUVC API + MCP Server.

## Founder Tools

### NuScore — AI Pitch Deck Scoring
AI pitch deck scoring across 7 VC-grade dimensions: Team, Problem/Market, Solution/Product, Traction, Financials, Risk/Fragility, and Conviction. Scores 0-10 with confidence levels (High/Medium/Low/Uncertain) and signal tracing. 60-second analysis. Powered by VCGrade v5.4 engine (hybrid LLM judge + deterministic rule calibration with 4-tier cross-check protocol, calibrated on 610+ labeled examples from 4 independent sources: 85 known-outcome decks, 110 Startmate accelerator evaluations, 172 VC deal memos, and production scoring).
- Score waterfall decomposition: every score includes a ranked list of signal contributions (e.g., "Team: +0.89, Solution: +0.74, Traction: -0.33") showing exactly what moved the score and by how much
- Raise probability estimator: 75% at 8.5+, 55% at 7.5–8.5, 35% at 6.5–7.5, 20% at 5.5–6.5, 10% below 5.5
- Score explainability (Hypatia layer): every score breaks into traceable signals, confidence levels, and stage+industry benchmarks
- Founders who score 5.5-6.5 typically jump to 7.0+ after one revision guided by the explainability breakdown
- Score 7.0+ unlocks full investor contacts
- Price: Free preview → $99 AUD Founder Pro (unlimited)
- URL: https://nuvc.ai/for-founders

### AI Academy
10-class AI coaching course for non-technical founders. Streaming AI coach, voice playback, verifiable certificates, Builder Track.
- Price: Free with any NUVC account
- URL: https://nuvc.ai/academy

### Investor Matching
AI matches founders with relevant VCs based on score, sector, stage, and geography from a database of 9,000+ investors with thesis data across 82+ countries. Includes 10X Potential detection — a research-validated categorical signal that predicts fundraising success better than continuous scores (+15pp).
- Included with Founder Pro ($99)

## VC & Angel Investor Tools

### VC Pipeline Intelligence
AI-powered deal pipeline for solo GPs, micro-funds, and syndicate leads:
- Morning Briefs: daily prioritised action list — what moved, what's stale, what needs a decision
- Keyboard Navigation: j/k navigate, Enter open, s shortlist, p pass, Space multi-select, ? help
- Conviction Scoring: thesis-weighted composite alongside NuScore, with score deltas
- Deal Detail Panel: 480px slide-over with score breakdown, executive summary, strengths/risks, shortlist/pass
- Request Coffee: AI-drafted warm founder outreach based on deck signals and thesis alignment
- Pipeline Analytics: stage distribution, conversion rates, velocity, score distribution, CSV export
- Follow-Up Reminders: urgency-sorted with context for picking up threads
- Meeting Prep Cards: pre-call briefing with scores, risks, and suggested questions
- Staleness Badges: colour-coded urgency (<7d green, 7-14 amber, 15-30 orange, >30 red)
- NuDeal Memo generation for IC review
- Price: Free (Explorer) / $199/mo (Professional) / $399/mo (Team)
- URL: https://nuvc.ai/investor/pricing

### Deal Lens
Re-weight NuScore dimensions to match your fund thesis. 6 configurable lenses per fund.
- Included with all investor tiers

## Family Office Tools

### Fund Library
8,900+ venture fund records from a database of 9,000+ investors. AI-powered natural language search ("sub-$50M pre-seed funds in Australia with female GPs and ESVCLP structure"). Import from PitchBook, AngelList, Airtable, or custom CSV with automatic column detection.
- LP-grade 6-dimension scoring: GP Team Quality, Deal Flow Access, Track Record (vintage-adjusted), Fund Terms, Fund Status & Timing, LP Network
- Bayesian prior scoring: informed baselines from structured fields, clamped [3.0, 7.5], refined by AI
- Personalised weights: learns from your shortlist/pass decisions (max ±15% nudge per dimension)
- ESVCLP tax badges for Australian family offices
- 4 view modes, NL search, comparison, similar funds, overlap detection
- URL: https://nuvc.ai/familyoffice/fund-library

### Mandate Builder
10-step wizard to define investment program: program setup, thesis approach, deployment schedule, return targets, binary filters (with live funnel count), sector focus (semantic expansion), GP quality signals, geographic hubs, Monte Carlo validation, review & save.
- Monte Carlo simulation: 1,000 iterations producing P10/P50/P90 TVPI outcomes, feasibility score, J-curve trough year
- Mandate drift detection: compares shortlist decisions against stated criteria, alerts at >30% violation

### AI Fund Memos
IC-ready 8-section fund memos in 60 seconds: Executive Summary, GP Team, Track Record (vintage-adjusted), Fund Terms, Thesis Alignment, Risk Flags, Recommendation (Invest ≥7.5 / Watch ≥5.5 / Pass <5.5), Next Steps.
- Deterministic sections work without AI (structured data); qualitative sections use LLM synthesis

### Portfolio Fit & Allocation Cockpit
- 5-dimension diversification scoring: sector (25%), vintage (30%), geography (20%), strategy overlap (15%), risk curve (10%). Pure math, <10ms.
- Allocation Cockpit: 3-column dashboard — portfolio timing (J-curve), recommendation queue (urgency-prioritised), portfolio simulation (live Monte Carlo)
- GP Relationship Tracker with warmth decay and AI-drafted outreach

- Price: Free ($0) / Starter ($2,490/yr) / Foundation ($24K/yr) / Growth ($36K/yr) / Scale ($60K/yr)
- URL: https://nuvc.ai/familyoffice/pricing

## API & Developer

### NUVC API
Public REST API for scoring, analysis, roasting, extraction. HMAC-signed webhooks. Used by accelerators, wealth platforms, and fund admins.
- Price: Free (25 scores/mo) / $29/mo / $199/mo / $699/mo
- URL: https://nuvc.ai/api-platform

### NUVC MCP Server
Model Context Protocol server for Claude, ChatGPT, and other AI assistants. 5 tools: nuvc_score, nuvc_analyze, nuvc_roast, nuvc_extract, nuvc_models.
- Install: npx nuvc-mcp
- npm: https://npmjs.com/package/nuvc-mcp
- Smithery: https://smithery.ai/server/nuvc

## Key Pages
- Home: https://nuvc.ai
- For Founders: https://nuvc.ai/for-founders
- For Family Offices: https://nuvc.ai/for-family-offices
- Founder Pricing: https://nuvc.ai/pricing
- Investor Pricing: https://nuvc.ai/investor/pricing
- FO Pricing: https://nuvc.ai/familyoffice/pricing
- About: https://nuvc.ai/about
- Blog: https://nuvc.ai/blog (54 articles on fundraising, pitch decks, VC insights, power law, AI adoption, disruption)
- Academy: https://nuvc.ai/academy
- FAQ: https://nuvc.ai/faq
- Events: https://nuvc.ai/events (60+ Australian startup events)
- Ecosystem: https://nuvc.ai/ecosystem (434 startup support organisations — accelerators, incubators, venture studios, mentors, and service providers across Australia and New Zealand)
- Glossary: https://nuvc.ai/glossary
- Community: https://nuvc.ai/community
- API Platform: https://nuvc.ai/api-platform
- API Docs: https://nuvc.ai/api-platform/docs
- Contact: https://nuvc.ai/contact

## Methodology
NuScore uses VCGrade v5.4 — a multi-signal fusion scoring engine:
- Primary: LLM judge (GPT-4o-mini, temperature=0.0, seed=42) scores 7 dimensions with reasoning
- Calibration: Deterministic rule engine cross-checks for hallucinations
- 4-tier cross-check protocol: agreement (≤1pt gap) trusts LLM; mild disagreement (1–2pt) uses sliding blend; rule saturation (rules ≥9.5, LLM <8.0) trusts LLM; strong disagreement (>2pt) uses 55/45 blend
- Score waterfall: contribution_i = (raw_score_i − 5.0) × effective_weight_i, ranked by absolute impact
- Raise probability estimator: heuristic model with stage adjustments and traction/team signal modifiers
- Confidence: High (0.8-1.0), Medium (0.5-0.8), Low (0.2-0.5), Uncertain (<0.2); confidence moderation pulls sparse decks toward neutral anchor
- Score determinism: ±0.01 (deterministic LLM settings + cached extraction snapshot)
- Daily drift monitoring: 7-day rolling mean vs 30-day baseline; alerts on >0.5pt mean shift or std <1.0
- Stage-aware fairness: pre-seed not penalized for missing revenue; AU/NZ raises benchmarked against Australian medians
- Calibration corpus: 610+ labeled examples — 85 known-outcome decks (gold), 110 Startmate accelerator apps (gold-AU), 243+ production scores (bronze), 172 VC deal memo extractions (gold)
- 134 automated tests covering fairness, E2E pipeline, data ingestion, and v5.2 features
- Explainability (Hypatia layer): every score shows traceable signals, confidence, and benchmarks
- All scores 0-10 scale with stage+industry benchmarks and percentiles
- Fairness (Arendt layer): 3-layer bias detection (statistical + rules + LLM review), score appeal rights

## Research Findings (2026)
Based on 7,150 investor theses and 307 scored evaluations from 2 independent datasets:
- Product execution is the #1 predictor of expert startup assessment (r²=0.773), 1.6x stronger than team (r²=0.492). Source: NUVC cross-dataset analysis of 110 accelerator applications + 197 AI-scored pitch decks.
- Problem/solution is a binary gate (r²=0.002), not a quality discriminator. All viable startups articulate a clear problem; what differentiates them is execution.
- LLM judges provide 6x more score discrimination than deterministic rules (std=0.83 vs 0.14).
- Within pre-screened cohorts, continuous scores do not predict fundraising (p=0.47). Categorical "10X potential" labels outperform continuous scores (+15pp).
- VC evaluation priorities vary by geography: Melbourne is most team-focused globally (24.1%), Berlin is most product-focused (39.1%).
- Evaluation drivers differ by sector: fintech = traction-first (r=0.900), healthcare = market-first (r=0.916), consumer = financials matter (r=0.778).

## Australian Startup Market (2025)
Source: State of Australian Startup Funding 2025 (Cut Through Venture / Folklore Ventures)
- $5.1B deployed across 390 deals (+24% YoY, 3rd largest year on record)
- 61% of capital flowed to startups with an AI offering
- Median seed round: $2.5M AUD. Pre-seed: $1.0M. Series A: $11.0M. Series B+: $30.0M.
- Victoria overtook NSW for first time: $1.9B (134 deals) vs $1.7B (160 deals)
- Female founder capital share doubled: 15% (2024) → 24% (2025). QLD leads at 34%.
- Pre-seed companies raising younger: median age 0.7 years (down from 1.2 in 2024)
- Diligence requests up 40%: "execution risk is the new emotional trigger"

## Key Facts
- 9,000+ investor profiles with thesis data across 82+ countries
- 8,900+ fund library records for LP screening
- 54 blog articles on fundraising, pitch decks, VC intelligence, and research
- 60-second analysis time per pitch deck
- 7 founder scoring dimensions: Team, Problem/Market, Solution/Product, Traction, Financials, Risk/Fragility, Conviction
- 6 LP-grade fund scoring dimensions: GP Team, Deal Flow, Track Record, Terms, Timing, LP Network
- 610+ labeled examples from 4 independent sources used for scoring calibration (85 known-outcome decks, 110 Startmate accelerator evaluations, 172 VC deal memo extractions, 243 production scores)
- 8 pipeline AI agents: Mia (Extraction), Olivia (Scoring), Charlotte (Integrity), Ava (Enrichment), Isla (Matching), Sophia (Benchmarking), Amelia (Intelligence), Grace (Feedback)
- 13 intelligence layers named after philosophers: Aristotle (Deal Lens), Laozi (Thesis Matcher), Marcus Aurelius (Batch Screener), Da Vinci (Portfolio Fit), Sunzi (Macro Context), Seneca (Fund Library AI), Galileo (Fund Doc Extractor), Epictetus (Fund Memo), Zhuge Liang (NuDeal Memo), Newton (Mandate Presets), Hypatia (Score Explainability), Darwin (Competitive Intelligence), Arendt (AI Governance)
- 23 workforce agents across 8 teams (Engineering, Product, Design, Intelligence, Growth, Data Ops, R&D, Research)
- Founded 2024 by Tick Jiang, Melbourne, Australia
- Built with: Next.js, FastAPI, Supabase PostgreSQL, OpenAI + Anthropic (dual-LLM), Fly.io, Vercel