Every dimension of
connection.

Serendipity is built on 27 psychological frameworks, zero-knowledge cryptography, and the belief that the best connections happen in person. Here is everything the app does -- in plain language and under the hood.

Discovery Understanding You The Encounter Safety & Privacy Intelligence Adaptive Experience

Discovery 3 features

How Serendipity finds the right people around you.

◎ Proximity Detection

The app detects compatible people within Bluetooth range -- no GPS, no location tracking. Your phone quietly listens for other Serendipity users nearby, and all matching happens on your device.

How it works
Uses BLE GATT advertising to broadcast a 41-byte fingerprint containing encrypted trait fragments. Beacon IDs rotate every 15 minutes using BLAKE2b hashing, making it impossible to track a user across sessions. Range is hardware-dependent, typically 10-200m depending on tier. No GPS or network location is ever accessed.
◆ Resonance Scoring

A compatibility score computed from your interests, skills, thinking style, values, and personality. It tells you not just who is nearby, but who is worth meeting.

How it works
A 6-dimension weighted vector (interests, skills, communication style, energy, problem alignment, optional synastry) is compressed into semantic bloom filters. Resonance is computed entirely on-device by comparing bloom filter intersections and cosine similarity across dimensions. No trait data is ever transmitted in the clear.
⬣ The Mesh Radar

A visual radar showing who is nearby and how strongly they resonate. Closer dots mean higher compatibility. The radar is your window into the invisible social field around you.

How it works
Rendered as an SVG radar with RSSI-based distance estimation. Signal strength is translated to approximate distance rings. Dot position encodes resonance score (radial) and dimensional category (angular). Animated pulse indicates active broadcasting. All rendering is local; no positional data is shared.

Understanding You 8 features

How the app learns who you are -- without ever asking.

◗ Behavioral Fingerprinting

The app learns who you are from how you use it -- not from forms you fill out. Your scroll speed, the words you linger on, the encounters you seek: these paint your portrait.

How it works
Captures 25+ behavioral metrics (session duration, interaction latency, content engagement patterns, transmission editing frequency, glyph selection tendencies) and maps them to Big Five personality traits, Schwartz Values, and attachment style inference. All computation is on-device. Raw behavioral data never leaves your phone.
★ Personality Model

Big Five personality, 10 core values, attachment style -- all inferred, never asked. No quizzes. No self-reporting bias. Just what your behavior reveals.

How it works
Weighted inference from behavioral signals (interaction patterns, timing, content choices) and linguistic signals (transmission wording, glyph selection). Produces continuous scores across Big Five (OCEAN), Schwartz 10 values, and attachment dimensions (anxiety/avoidance). Confidence intervals widen when data is sparse, narrow with more encounters.
⚙ Cognitive Resonance

Matches how you THINK, not just what you think about. Two people who both like philosophy but think in radically different ways will produce a different resonance than two who think alike.

How it works
Maps users along multiple cognitive axes: convergent vs. divergent thinking, abstract vs. concrete orientation, integrative complexity (tolerance for nuance), epistemic style (empirical vs. intuitive), and Jungian cognitive functions (Ti/Te/Fi/Fe/Ni/Ne/Si/Se). Compatibility can reward similarity OR complementarity depending on the axis.
♥ Emotional Profile

Sensitivity type, regulation style, bandwidth -- your nervous system's social signature. The app understands how much social energy you have and matches accordingly.

How it works
Builds a profile using polyvagal mapping (ventral vagal / sympathetic / dorsal vagal tendencies inferred from engagement patterns), HSP detection (highly sensitive person markers from session behavior), and social metabolic rate (how quickly social energy depletes, measured by session tapering and encounter spacing).
✎ Narrative Identity

Where you are in your life story -- and who resonates with your chapter. People in a reinvention phase connect differently than people in a building phase.

How it works
Applies McAdams' narrative identity theory to detect redemptive vs. contamination narrative arcs from transmission content over time. Maps to Erikson developmental stages (identity vs. role confusion, intimacy vs. isolation, generativity vs. stagnation). Life chapter matching pairs users whose current chapters are complementary or resonant.
⚒ Values Archaeology

Discovers what you truly value from how you behave, not what you claim. What you do when no one is measuring reveals more than any survey.

How it works
Analyzes exchange orientation (communal vs. transactional encounter patterns), problem hierarchy (which transmission topics recur under stress), perception framework (deficit vs. abundance framing in language), and maps all signals to the Schwartz circumplex (10 values arranged by motivational compatibility and conflict).
◑ Shadow Profile

What others experience of you vs. what you project. Hidden strengths you did not know you had. The gap between your self-image and your social impression, illuminated gently.

How it works
Compares your self-model (derived from your behavior and transmissions) against the glyph-derived perception model (aggregated from glyphs others leave about you). Archetype gap analysis identifies where self-image diverges from social impression. Blind spot reframing presents these gaps as undiscovered strengths rather than deficiencies.
↗ Growth Trajectory

Where you are heading, not just where you are. The app detects your direction of growth and matches you with people who catalyze that trajectory.

How it works
Runs transmission history regression to detect topic drift over time. Semantic cluster momentum identifies which interest/skill domains are accelerating or decelerating. Inflection point detection recognizes when a user is about to shift into a new phase -- and matches them with people who have recently completed similar transitions.

The Encounter 7 features

What happens when two people decide to meet.

✋ The Handshake

Mutual consent at every step. Nothing is revealed until both people agree. You see a resonance score first. Then you both choose whether to reveal transmissions. Then you meet.

How it works
Implements a 4-state protocol: PING (anonymous resonance notification) → ACK (mutual interest confirmed) → REVEAL (encrypted transmissions exchanged) → ENCOUNTER (in-person meeting begins). Each transition requires explicit consent from both parties. State machine enforces that no data flows backward.
◉ Progressive Depth Layers

The more you meet someone, the deeper the app reveals. First encounters show surface compatibility. Repeated encounters unlock character insights, shadow comparisons, and full resonance history.

How it works
5-layer gating system: Surface (composite score, basic interests) → Context (skill gaps, mutual problems) → Character (values alignment, cognitive style match) → Shadow (archetype gap, blind spot complements) → Resonance History (full trajectory, encounter quality over time). Each layer unlocks after a threshold number of quality encounters.
❖ Connection Archetypes

The app names what your connection IS: Intellectual Sparring Partners, Complementary Builders, Kindred Spirits, Catalyst and Student, Parallel Journeyers, and more.

How it works
12 archetypes are determined by score pattern matching across all resonance dimensions. For example, high cognitive resonance + low interest overlap = "Intellectual Sparring Partners." High values alignment + high skill complementarity = "Complementary Builders." Archetype is computed per-pair and can evolve over repeated encounters.
★ Encounter Quality Prediction

Not just "are you compatible" but "will THIS encounter be good?" The app considers time of day, your energy level, how long since you last met, and dozens of contextual factors.

How it works
Combines contextual factors: time of day (mapped to both users' energy curves), zone type (cafe vs. transit vs. event), fatigue estimation (session count today), relationship depth (encounter history), and chronotype alignment (are both users in their peak window?). Produces a 0-1 encounter quality prediction that supplements the static resonance score.
⚡ Conversation Primer

Never starts with "so what do you do?" The app generates specific spark points based on what makes your connection unique. Not small talk. Real talk, from the first word.

How it works
Performs cross-match analysis to identify the most unusual overlaps and complements between two users. Skill gap detection finds where one person has expertise the other is seeking. Template selection chooses from context-appropriate conversation frameworks (exploratory, collaborative, philosophical, practical) and generates 2-3 specific openers.
▼ Conversational Depth Engine

Detects how deep your conversation is going, in real time. If you are both going deep, the app stays quiet. If the conversation is stuck at surface level, it offers a nudge.

How it works
5-level NLP classification of conversational depth (phatic → factual → evaluative → gut-level → peak). Asymmetry detection identifies when one person is going deeper than the other. Depth-appropriate suggestions are offered only when both users are at the same level and could benefit from a prompt to go further.
✦ Glyphs

One word. That is all you leave. After an encounter, you choose a single glyph -- brilliant, catalytic, grounding, enigmatic -- that captures the essence. Accumulated over encounters, they form your aura.

How it works
30 constrained vocabulary words organized into 5 categories (intellectual, energetic, emotional, creative, relational). Each glyph maps to VIA Character Strengths for psychometric validity. Glyphs accumulate into a weighted distribution that forms the user's public "aura" -- visible to future encounters. Cannot be gamed because each encounter yields exactly one glyph from each participant.

Safety & Privacy 6 features

How the app protects you -- architecturally, not just by policy.

🔒 Zero-Knowledge Architecture

Not even we know who you are or where you have been. This is not a privacy policy -- it is a mathematical guarantee. The system is designed so that tracking you is architecturally impossible.

How it works
X25519 key exchange for ephemeral session keys. XSalsa20-Poly1305 authenticated encryption for all data in transit. BLAKE2b hashing for beacon IDs with 15-minute rotation. No persistent identifiers in the protocol layer. The server relays encrypted blobs it cannot decrypt. Location data never leaves the device.
👥 Buddy Beacon

Link a friend. They get a silent alert when your encounter starts. If anything feels off, your buddy already knows your approximate area. Peace of mind without compromising your independence.

How it works
Sends a push notification to a pre-linked device when an encounter enters the REVEAL state. Notification includes only a zone geohash (neighborhood-level, not precise location) and a timestamp. The buddy can trigger a check-in prompt or escalate to emergency services. All buddy links are encrypted and revocable.
⚪ Encounter Boundaries

Set your terms before you meet. Both users see each other's boundaries upfront: time limits, conversation topics, meeting context. No surprises. No pressure.

How it works
Structured boundary display presents both users' preferences in a normalized, non-judgmental format. Boundaries are categorized (time, space, topic, intent) and presented symmetrically so neither party feels they are the "restrictive" one. Normalized presentation ensures that setting boundaries feels like a natural part of every encounter, not an exception.
✓ Safety Check-Ins

15 minutes in, the app asks: "Everything good?" One tap. If you do not respond, your buddy gets an alert. Unobtrusive. Reliable. Always there.

How it works
A scheduled local notification fires at configurable intervals (default 15 minutes) during an active encounter. Responding "all good" resets the timer. Escalation on no-response: if no acknowledgment within 3 minutes, a secondary notification is sent to the linked buddy device. The check-in is disguised as a generic notification to avoid social awkwardness.
★ Trust Scoring

Built from real encounters and glyphs -- cannot be bought or faked. Your trust score grows only through genuine, positive in-person encounters over time.

How it works
Composite of encounter count (volume of real meetings), glyph count and distribution (quality and variety of impressions left by others), and safety report aggregation (absence of negative flags). Score is time-weighted (recent encounters matter more) and cannot be inflated by repeated encounters with the same person.
🔑 Secure Storage

Sensitive data encrypted on-device. Different security levels for different data types. Your personality model and encounter history are stored with the same rigor as your banking app.

How it works
Uses expo-secure-store for hardware-backed keychain/keystore access. Data is classified by sensitivity level (public transmission vs. personality model vs. encounter history vs. cryptographic keys) with each level receiving appropriate encryption and access controls. TTL enforcement automatically purges expired data -- encounter details expire, glyphs persist.

Intelligence 7 features

The deeper patterns the mesh sees that you cannot.

✪ Synchronicity Scoring

How improbable was this specific encounter? The app measures meaningful coincidence. When the right person appears at the right time in the right place, the synchronicity score spikes.

How it works
Computes the product of four improbability signals: tag rarity (how uncommon your shared interests are in this zone), zone novelty (are you both outside your normal areas?), temporal improbability (outside your usual schedules?), and resonance score. Higher synchronicity = more statistically unlikely the encounter was random chance.
✸ Weak Tie Detection

Tells you when someone from OUTSIDE your bubble is nearby. The most transformative connections often come from people who think nothing like your current circle.

How it works
Based on Granovetter's weak tie theory. Performs network diversity analysis by comparing incoming resonance profiles against your existing encounter history. Computes a bridging coefficient: how different is this person's profile from everyone you have met before? High bridging coefficient triggers a special "outside your bubble" indicator.
⚛ Generative Potential

Predicts what you could CREATE together. Not just "you are compatible" but "together, you could build something neither of you could build alone."

How it works
Analyzes skill complementarity gaps (your skills + their skills = complete project capability), problem adjacency (you are working on related but non-identical problems), domain bridges (expertise in different fields that cross-pollinate), and cognitive catalyst detection (thinking styles that amplify each other rather than duplicate).
∞ Self-Expansion Scoring

Will knowing this person change who you are? The app predicts whether an encounter will expand your identity, skills, or worldview in ways you did not expect.

How it works
Based on Aron's self-expansion model. Scores three dimensions: novel domain exposure (topics and skills you have never encountered), skill acquisition potential (can you learn from this person?), and identity expansion prediction (will this connection change how you see yourself?). High self-expansion scores indicate transformative rather than merely compatible encounters.
↑ Broaden-and-Build Tracking

Tracks whether your encounters are creating an upward spiral. Are your connections making you more open, more curious, more capable? Or are they stagnating?

How it works
Implements Fredrickson's broaden-and-build theory. Tracks encounter quality cascading: does a good encounter lead to more openness in the next? Measures topic diversity growth, encounter frequency trends, and glyph sentiment trajectory. Trajectory prediction identifies whether you are in an upward spiral, plateau, or contraction phase.
◈ Mesh Insights

The oracle speaks: patterns in your encounters you would never notice. "You connect best with builders on Tuesday evenings." "Your strongest encounters happen outside your home zone."

How it works
Resonance archaeology mines your encounter history across 11 insight types: temporal patterns, spatial patterns, dimensional strengths, archetype frequencies, glyph clusters, depth progression rates, energy correlations, topic evolution, network structure, synchronicity trends, and growth trajectory inflections. Insights are generated locally and presented as periodic "oracle" messages.
★ Federated Learning

The mesh gets smarter without anyone's data being exposed. The system learns from aggregate patterns while mathematically guaranteeing that no individual's data can be extracted.

How it works
Uses differential privacy with ε=1.0 (strong privacy guarantee). Each device contributes only aggregate model updates, never raw data. Laplacian noise is added to all contributions before they leave the device. The central model improves from aggregate-only contributions, making it mathematically impossible to reverse-engineer any individual's data from the shared model.

Adaptive Experience 5 features

An app that reshapes itself around who you are.

▦ Adaptive Interface

The app reshapes itself around your personality. Introverts see a quieter, less information-dense interface. High-openness users see more experimental layouts. The app meets you where you are.

How it works
Adjusts four interface parameters based on personality model: information density (how much data is shown at once), visual intensity (animation frequency, color saturation), notification cadence (how often and how urgently the app interrupts), and beacon sort order (what criteria determines which nearby people appear first). All personality-driven, all adjustable manually.
☼ Chronotype Matching

Morning person? Night owl? The app knows and matches accordingly. No more trying to schedule a coffee with someone whose brain does not turn on until 10pm.

How it works
Circadian inference from session timing patterns (when you open the app, when you broadcast, when you accept encounters). Classifies into 4 chronotypes (lion/bear/wolf/dolphin). Computes energy curve overlap: what percentage of each user's peak hours align? Factors into encounter quality prediction so the app suggests meetings when both users are energized.
✨ Neurodivergent-Aware

Your brain is wired differently? The app honors that. ADHD, autism spectrum, HSP -- the matching engine recalibrates so that neurodivergent traits are assets, not noise.

How it works
Big Five recalibration adjusts personality scoring to account for neurodivergent baselines (e.g., low agreeableness in autism spectrum is not the same as low agreeableness in neurotypical profiles). ND-specific matching bonuses reward connections where neurodivergent traits complement (e.g., ADHD divergent thinking + autistic systematic thinking). Communication tips are generated per-pair to bridge different processing styles.
◯ Holding Environment

The app measures whether a relationship is psychologically safe. Are both people reciprocating? Are boundaries being respected? Is depth growing organically?

How it works
Grounded in Winnicott's holding environment concept. Tracks three signals: reciprocity (are both users investing similar energy?), boundary respect (are stated boundaries being honored across encounters?), and organic depth scoring (is conversational depth increasing naturally, not being forced?). Low holding environment scores trigger gentle suggestions rather than penalties.
↻ Relationship Continuity

Connections do not die after the chat expires. The app keeps relationships alive through subtle, non-intrusive mechanisms that honor the organic rhythm of human connection.

How it works
Four continuity mechanisms: resonance pulses (periodic notifications when a previous encounter is nearby again), seasonal transmissions (quarterly prompts to update your broadcast, which re-surfaces relevant past connections), glyph evolution (your aura changes over time, triggering re-matching), and constellation nurturing (the app identifies clusters of connected people and facilitates group re-encounters).

36 features. One intention.

Real connection between real people, protected by real cryptography. Join the mesh.

Join the Waitlist Back to Home