delivery

HeyPBJ — Structured Matching Platform

Dating and social discovery as a routing problem—using cohorts, compatibility metrics, and guided prompts instead of endless feeds.

The Core Idea

Dating and social discovery aren’t content problems—they’re routing problems. HeyPBJ treats connection as infrastructure: users signal what they’re looking for, the system assembles compatible cohorts, and matches unfold through guided prompts rather than infinite scrolling.

Instead of feeds, swipes, and algorithmic mystery, HeyPBJ uses:

Declared preferences (what you’re seeking, how you want to interact)
Weighted signals (compatibility metrics across multiple dimensions)
Small cohorts (“rundles”—compatible groups of 3-8 people)
Guided prompts (structured interaction contexts)
Mutual visibility (everyone sees the same compatibility data)

The Problem with Feeds

Traditional dating/social apps treat discovery as an engagement optimization problem:

Endless swiping creates decision fatigue
Algorithmic feeds prioritize retention over connection
Matches lack context or shared understanding
Success metrics (time-on-app, swipes) misalign with user goals (meaningful connection)

The result: people spend hours scrolling but rarely find compatible matches, and when they do, there’s no shared framework for moving forward.

How HeyPBJ Works

1. Declared Preferences

Users explicitly state what they’re looking for:

Connection type (dating, friendship, collaboration)
Interaction preferences (pacing, communication style)
Values and priorities
Deal-breakers and must-haves

This isn’t a personality quiz—it’s routing metadata. The system uses these signals to compute compatibility.

2. Weighted Compatibility Metrics

Rather than binary yes/no, HeyPBJ calculates multi-dimensional compatibility:

Alignment on core values
Complementary interaction styles
Shared interests and context
Geographic/logistical feasibility

Scores are transparent and mutual: everyone in a rundle sees the same data about why they were matched.

3. Rundles (Small Cohorts)

Instead of one-on-one matches, the system creates small groups of 3-8 compatible people. This:

Reduces pressure of binary yes/no decisions
Provides social context (how you interact in a group matters)
Allows for organic subgroup formation
Creates shared understanding before pairing off

4. Guided Prompts & Decision Mechanics

Interaction happens through structured prompts rather than freeform chat:

Icebreaker questions
Group activities or challenges
Progressive disclosure (reveal info as trust builds)
Lightweight decision points (signal interest without commitment)

This pacing prevents premature intimacy and provides scaffolding for connection.

Design Principles

Intentionality Over Engagement

HeyPBJ optimizes for successful connection, not time-on-app. Users should spend less time in the system if it’s working.

Constraint-Based Matching

Small cohorts, explicit criteria, and pacing constraints make the space more legible. Instead of infinite choice, users get good options with clear rationale.

Mutual Visibility

No hidden algorithms. Everyone sees:

Why they were matched
Compatibility scores and dimensions
Who else is in their rundle
How the system made decisions

Context Before Chemistry

Compatibility data and group interaction provide shared context before one-on-one pairing. This reduces awkward cold-start conversations.

Technical Architecture

Frontend:

Progressive web app (PWA) for cross-platform access
Client-side compatibility scoring for privacy
Real-time interaction via WebSocket/AT Protocol events

Backend (AT Protocol-based):

Decentralized identity (users own their data)
Preference declarations as structured records
Compatibility computation as composable service
Rundle formation as routing layer
Interaction history stays with user, not platform

Matching Engine:

Multi-dimensional compatibility scoring
Constraint satisfaction for cohort formation
Pacing and decision mechanics
Progressive disclosure rules

User Journey

Onboarding: User declares preferences, priorities, and interaction style
Matching: System computes compatibility and forms rundles
Introduction: User receives cohort with transparency about why
Interaction: Guided prompts and group activities provide context
Decision Points: Lightweight signals to express interest in subgroups/individuals
Pairing: System suggests one-on-one connections within cohort based on mutual signals
Iteration: User refines preferences based on what worked

What Makes This Different

Not a Feed

No infinite scroll, no engagement maximization. Just periodic cohort assignments with clear rationale.

Not Algorithmic Mystery

Users see exactly why they were matched and can adjust their preferences to improve future rundles.

Not One-on-One from Start

Group context reduces pressure and provides richer compatibility data before pairing.

Not Platform Lock-In

AT Protocol foundation means users own their data and can move between compatible services.

Research Questions

HeyPBJ explores:

Can constraint-based matching produce better outcomes than infinite choice?
Do small-group dynamics provide better compatibility signals than profiles alone?
Does transparency (showing the algorithm’s work) increase trust and satisfaction?
Can pacing mechanisms prevent burnout and premature ghosting?
How do portable identity and preferences change platform power dynamics?

Part of the AT Protocol Suite

HeyPBJ is one of several interconnected projects exploring decentralized infrastructure:

AT Protocol Projects (Parent project)
Glowrm (Trust and reputation layer)
Leafroll (Professional identity)
Occupant (Public data infrastructure)

Glowrm integration: HeyPBJ uses Glowrm’s trust primitives for safety/moderation—users can carry reputation and verified credentials across networks.

Leafroll integration: Professional discovery rundles can pull from Leafroll profiles for skill-based matching and collaboration.

What This Demonstrates

Routing infrastructure as social product: Treating discovery as a systems problem rather than a content problem
Constraint-based design: How limits and structure can produce more humane outcomes than infinite choice
Transparent algorithms: Showing users why decisions were made builds trust and improves outcomes
Decentralized matching: Proving you can build discovery infrastructure without platform lock-in
Small-group dynamics: Using cohorts to provide context before one-on-one connection
Intentionality-first UX: Designing for successful connection rather than engagement metrics