The Science Behind the Platform
Your Network Is
Your Fundraise
We didn't build another CRM. We built on decades of network science to turn relationship chaos into your competitive edge.
↓ Explore the science“The warm introduction you need is closer than you think.”Built on 9 peer-reviewed papers across network science, VC research, and AI
Network Science
The Research That Drives Our Design
Our approach is grounded in decades of peer-reviewed research. Here are the four ideas that matter most.
The Strength of Weak Ties
Your acquaintances — not your close friends — open access to entirely new investor clusters. We map your entire team's network, not just the close relationships.
Granovetter (1973), Burt (1992)Small-World Navigation
Any VC is reachable in 2–4 introductions through shortcut edges. Our pathfinding models the distance-reducing potential of each hop, not just abstract shortest paths.
Watts & Strogatz (1998), Kleinberg (2000)Network Position Predicts Success
An investor's network position accounts for 67% of predictive power in outcome models. Companies backed by top-networked VCs have 2× lower failure rates.
Hochberg (2007), Bubna (2020)Three Networks, One Picture
Co-investment, affiliation, and social networks are distinct layers with independent signals. Most tools only capture one. We model all three.
Shi (2025), Poole (2025)Want the full research? Dive deeper.
Institutional Memory
Knowledge That Compounds
Fundraising is episodic — relationships are continuous. Your platform remembers what you've learned, strengthens what you use, and lets go of what you don't.
Relationships That Strengthen
Connections follow Hebbian learning. When two entities co-appear in news, co-invest, or are mentioned together, their connection strengthens. Active relationships naturally rise to prominence — round N+1 inherits the full signal of every prior round.
Intelligent Decay
Unused connections decay — but not by simple time. A connection that was exercised weekly and goes silent for a month fades faster than one that was always quarterly. Decay is relative to expected activity, so your institutional memory stays honest.
The Full Picture
Co-investor cliques form communities. Investor communities form clusters. Bridge nodes and weak ties connect the clusters. Your platform maps it all — and between the filaments of activity, it finds the voids where new bridges create the most value.
Community Intelligence
Smarter Together, Private Always
Every founder who uses CastleFort makes the platform smarter for everyone — without ever exposing private deal data.
Canonical + Overlay
A shared knowledge commons grows with every user. Your private annotations, notes, and deal context stay entirely yours. The canonical layer improves; your overlay remains invisible.
Anonymous by Default
When you research an investor, verify a connection, or correct a data point, that knowledge flows back anonymously. Individual contributions are never attributable; collective intelligence compounds.
Multi-Signal Scoring
No single source is trusted alone. Community-contributed data, public records, and AI extraction are weighted independently. Confidence increases with source diversity, not volume.
How We Build
Four Design Principles
Architectural choices that shape what the platform can do and how much you can trust it.
Epistemic Humility
Every data point carries a confidence score with explained provenance. The UI never lets you mistake speculation for fact.
Absence Taxonomy
Missing data isn't a black box. Every gap is classified: undiscovered, private, nonexistent, or expired.
Entity Resolution
The same person across CRM, news, and SEC filings — resolved with quantified confidence.
Privacy by Architecture
Multi-tenant isolation, local AI processing, and zero-knowledge defaults. Security is structural, not a policy checkbox.