FoundersSuite

Beall-Butterworth

B2B discovery marketplace matching early-stage founders with domain-credible end users—then turning conversations into investor-ready insight.

Role: Beall-Butterworth
Proof: Stella Zhang NVC

Problem

Founders need real B2B signal before MVP, but cold outreach converts poorly and warm networks skew positive. In our research, all 20 founders interviewed still used personal networks for validation despite knowing the bias—the Mom Test failure mode in practice.

Architecture

Match → converse → synthesize

Founder                 Matcher Service              Tester
   │                          │                         │
   │  POST /match             │                         │
   │  {domain, needs}         │                         │
   │─────────────────────────▶│  SVD(founder, testers)  │
   │                          │  rank by experience     │
   │◀──── ranked feed ────────│                         │
   │                          │                         │
   │  accept candidate        │  invite                 │
   │─────────────────────────▶│────────────────────────▶│
   │                          │◀──── accept / decline ──│
   │                          │                         │
   │◀──── booked session ─────│                         │
   │                          │                         │
   │  conversation done       │  LangChain agents       │
   │─────────────────────────▶│  extract + bias check   │
   │◀──── insight summary ────│                         │

Component view

┌─────────────────────────────────────────────────────┐
│                    React SPA                        │
│  Founder dash │ Matcher feed │ Tester feed │ Board  │
└─────────────┬───────────────────────┬───────────────┘
              │                       │
              ▼                       ▼
     ┌─────────────────┐     ┌─────────────────┐
     │  SQL / profiles │     │  SVD engine     │
     │  bookings       │     │  domain weights │
     └────────┬────────┘     └────────┬────────┘
              │                       │
              └───────────┬───────────┘
                          ▼
                 ┌─────────────────┐
                 │  LangChain      │
                 │  insight + bias │
                 └─────────────────┘

Stack

React
TypeScript
Tailwind CSS
SQL backend
SVD matching
Fine-tuned LLM

What we built

SVD recommender over lived experience + problem context (MedTech-weighted), not demographics alone
Two-sided React app: onboarding, swipe-style review, community board
Post-round LangChain pipeline: theme extraction, WTP signals, leading-question bias nudges
Demand: 85% would use it; 65% called existing tools unreliable

Outcome

Working two-sided flow for Beall Butterworth; same narrative carried into Stella Zhang NVC. The product lives or dies on match quality—panels are easy to copy; domain-tuned experience graphs are not.