GaiaWM
World Infrastructure for AI
The first simulation-intelligence platform for fictional worlds —
from narrative perception to causal influence modeling
gaiawm.github.io | openhistorymap.org | fantasymaps.org | rpg-schema.org
📂 Data room — technology, runway, business model
The Problem: AI Agents Are Lost
- LLMs process text but lack spatial, temporal, and social context
- RAG over wikis → Hallucinations, unstructured, no reasoning
- Hand-crafted game AI → Expensive, brittle, doesn't scale
- Current "world models" → Static maps, no causal simulation
Result:
$XXB invested in AI agents that fail because they don't understand environments or predict consequences
Three Live Products, One Engine
Most AI infrastructure pitches vaporware. We ship.
👁️
EyeOnWorld
Narrative perception from map data. Click any location, get immersive description.
LIVE
openfantasymaps.org
🔍
Object Inspector
Deep lore examination with canonical sourcing. Verify any object against wiki canon.
LIVE
gaiawm.github.io/inspector
⚙️
GaiaWM API
Full influence engine + agent tracking. REST API + MCP organ servers live.
LIVE
MCP LIVE
"Looking around, you're standing at the edge of a residential neighborhood where the midday sun beats down on the cobblestones. Selduth Street is just a few paces to your right..."
— EyeOnWorld generated from Waterdeep map data
The Solution: Simulation Intelligence
GaiaWM = Belief-centric world models + Multi-channel influence simulation
🗺️ Spatial Layer
40+ worlds, GIS precision
Toril, Star Wars, Middle Earth
⏰ Temporal Layer
Time-aware state
History, seasons, cycles
👥 Social Layer
Factions, relationships
Harpers, Zhentarim, reputation
⚙️ Operational Layer
Markets, quests, events
Dynamic world state
🌊 Influence Layer
Causal simulation
7 channels, 10 mechanisms
🧠 Narrative Layer
Perception & description
EyeOnWorld generation
Product → API Architecture
Free tools validate demand. API enables scale.
USER EXPERIENCE LAYER
EyeOnWorld • Object Inspector • GaiaWM API
↓
GAIAWM INFLUENCE ENGINE
Multi-channel analysis • Scenario simulation • Ripple propagation
↓
FANTASYMAPS WORLD DATA
40+ canonical worlds • Wiki-linked • Copyright-compliant
Same infrastructure, three interfaces: Consumer (free) → Power user (free) → Developer (paid)
The Influence Engine: Our Moat
No other platform simulates how events propagate through complex environments.
7 Influence Channels
- • Physical (weather, terrain)
- • Social (rumor, reputation)
- • Economic (trade, prices)
- • Informational (news, secrets)
- • Military (force, patrols)
- • Symbolic (law, culture)
- • Temporal (history, cycles)
10 Propagation Mechanisms
- • Travel, Rumor, Law
- • Fear, Force, Signal
- • Line of sight, Memory
- • Weather, Trade
Example: Fire in Dock Ward → spreads via weather (physical) + rumor (social)
POST /influence/ripple
Simulate cascade effects over time
POST /influence/briefing
AI-generated strategic intelligence
POST /influence/explain
Interpretable causality (scientific/bard/strategist)
Traction: 5 Years, 4 Projects, 1 Methodology
2016
OpenHistoryMap
ArcheoFoss (2021-2025), digital humanities adoption • Methodology foundation
2022-23, 2026
FantasyMaps + RPG-Schema
40+ worlds documented, 10K+ users, community growth • Data infrastructure
2024
Rivoluzioni Tangibili
40+ player participatory simulation • Multi-agent validation
2025
SciTePress Paper + Live Products
Peer-reviewed industrial digital twins + EyeOnWorld/Inspector/API live
2026
Embodied Agents + Ownership Layer
"The Ghost in the Shell" talk (July) • 8-organ agent runtime live • ghostkit + GhostDeck (BYOK ownership) • 10K+ calibration records
The SciTePress Validation
Industrial Proof, Academic Rigor
"Digital Twin with an Agentic Approach on a Multi-Layer Graph"
— SciTePress 2025
- Multi-layer graphs enable agentic reasoning over complex systems
- Real-time monitoring + predictive simulation in industrial infrastructure
- Architecture scales from Bologna's water supply to Baldur's Gate's economy
- Now implemented, not just published: the engine executes the paper's rules language — the paper's own listings run as its conformance tests, calibrated live against Bologna's open traffic data
Why this matters for VCs:
- Not "research" — production-validated methodology
- Risk reduction: works in safety-critical industrial contexts
- Team can execute: published, peer-reviewed, reproducible
Market Opportunity: Four Converging Waves
| Market |
Size |
Our Position |
| AI Agent Infrastructure |
$7B, $180B by 2030 |
Only simulation-intelligence platform |
| Game/VTT Middleware |
$5B, $13B by 2030 |
First academically-validated engine |
| Digital Twin Simulation |
$20B, $200B by 2030 |
Published methodology, proven team |
| Simulation Intelligence (NEW) |
Emerging |
Among the first movers, no competitors |
Convergence: Systems need training environments that are
Structured
Grounded
Dynamic
Causal
Beyond Games: Real-World Verticals
Same engine, real geography. Only the data plane changes: which world loads, which rules govern it, who the minds are.
🏙️ Urban Digital Twins
A twin that is populated and falsifiable — not a dashboard.
- • Ingests the city's open sensor data as ground truth (Bologna pilot live)
- • Measures its own divergence per rule — it told us "the model has no morning rush"
- • Counterfactuals: fork the city, inject the intervention, compare seeded runs
Municipalities, mobility, utilities — via consultancies & EU projects
🛡️ Defence & Disaster
Exercises where the civilian population actually behaves.
- • Agents perceive locally; rumor & stress propagate agent-to-agent
- • Per-side fog of war, injects as auditable events, courses of action as forked timelines
- • Bitemporal world → the after-action review is a query: who believed what, when
Civil protection first; defence via primes, under published dual-use guardrails
🏛️ Analysis of the Absence
The academic branch: the phenomenon is visible, its context unexplored.
- • Rebuild the period world from OpenHistoryMap — geometry + timeline
- • Populate with period-constrained embodied agents
- • Search for contexts whose dynamics reproduce the visible record
Deliberately non-revenue: grants, publications, calibration datasets
Business Model: One Engine, Three Markets
Not three products — three go-to-markets over one maintained engine. Every engineering week lands in all three at once.
🎮 Games
Freemium → Licence
Free indie → world licence → per-NPC usage → enterprise BYOK
- Buyer: studio CTO / tech director
- Alternative: 18 months of in-house platform work
- Free tools + OSS floor as the funnel
🏙️ Urban Twins
Build → Run → Ask
Integration engagement → annual calibrated-twin licence → per-study counterfactuals
- The calibration report is the SLA — we tell you where the twin is wrong, measured
- Channel: engineering consultancies
- EU projects fund the build phase
🛡️ Defence & Disaster
$50K-$500K
Per-exercise packages + standing readiness licence
- Sovereignty by default: on-prem, air-gappable, BYOK
- Entry via civil protection; defence via primes
- Published dual-use guardrails
We never monetise inference — agents run on the customer's model keys (BYOK). No provider lock-in, no token-price exposure, and it passes the sovereignty reviews others fail.
Full model in the data room →
Competitive Moat
| Competitor |
Their Approach |
Our Advantage |
| RAG vendors |
Vector search over text |
Structured graphs + causal simulation |
| Procedural gen |
Random generation |
Canonical fidelity, community-verified |
| Game engines |
Hand-crafted AI |
Belief-centric, multi-agent, scalable |
| AI world models |
Tech demos, no live products |
3 live products + 10 years production |
1
Methodology Depth
10 years, peer-reviewed, conference-validated
2
Data Network Effects
40+ worlds, cumulative documentation — plus a calibration dataset scored against measured reality (Bologna sensors), unforgeable by a fast follower
3
Simulation Intelligence
No competitor has causal influence engine
Go-to-Market: Land and Expand
Now
Phase 1: Product-Led Growth
EyeOnWorld/Inspector drive awareness • API early access • Target: 100+ dev signups
M4-9
Phase 2: Vertical Expansion
Gaming: Foundry VTT partnership • Enterprise: First digital twin pilot • MCP launch
M10-18
Phase 3: Platform Scale
World simulation marketplace • "GaiaWM Verified" certification • Research grants
The Team
👤 Marco Montanari
Co-founder / CTO
- 10 years world infrastructure
- OpenHistoryMap & FantasyMaps author
- SciTePress 2025 (industrial DT) • ArcheoFoss track (2021-2025)
👤 Lorenzo Gigli
Co-founder / Architect
- PhD
- Blockchain & AI learning models
👤 Matteo Sipione
Co-founder / Architect
- Narrative generative-AI models
- MCP/protocol development
Hiring: COO / commercial lead — enterprise sales, gaming industry
Advisory Board (in discussion):
Industrial DT experts (SciTePress co-authors) • AI agent researcher • Gaming veteran
The Ask
Series [A/Seed] Investment
$1.5M
💻 Engineering (50%)
MCP scaling, synthetic APIs, GaiaWM optimization, tooling and 3rd party developer experience facilitation
🤝 Sales/BD (25%)
Enterprise customers, VTT partnerships
📚 Research (25%)
Academic collaborations, conferences, funds for archaeology researches and worldbuilding initiatives to produce new datasets and new content
18-Month Milestones:
- 10 paying enterprise customers or $1M ARR
- 3 major VTT/platform partnerships
- 1000+ active API users, MCP server registry dominance
"We document worlds. Real, historical, fictional. We preserve their geography, structure their narratives, and enable intelligent systems to understand them. From Bologna's archaeology to Waterdeep's taverns to tomorrow's autonomous agents — context is everything."
GaiaWM
The simulation intelligence layer for contextual AI