Nine layers of intelligence turn a Bluetooth signal into a meaningful human encounter. Here is every step of the pipeline, from the first radio pulse to the Shadow Mirror.
Your phone continuously broadcasts a Bluetooth Low Energy beacon containing a rotating cryptographic identifier. No name, no photo, no location coordinates are ever attached to this signal. It is a pure presence pulse that says nothing more than "a Serendipity user is nearby."
Other devices running Serendipity detect this beacon within a range of 10 to 200 meters, depending on your subscription tier and power settings. The scanning happens in the background with minimal battery impact, using the same BLE advertising channels that fitness trackers and AirTags use.
Before any meaningful data is exchanged, your device performs a lightning-fast pre-filter using a compact 34-byte fingerprint embedded in the beacon payload. This fingerprint contains three components: a Bloom filter encoding your top interest categories, a seek/offer bitmap indicating what you are looking for and what you can provide, and an energy compatibility byte reflecting your current interaction mode.
The Bloom filter checks for interest overlap in constant time. The seek/offer bitmaps are cross-matched so your "seeking" bits align with their "offering" bits and vice versa. Devices that fall below the minimum threshold are silently discarded before any handshake occurs.
Once a device passes the pre-filter, an encrypted handshake establishes a temporary session. Both devices exchange their full transmission payloads and compute a multi-dimensional resonance score entirely on-device. No server is involved. Six dimensions are evaluated:
Interest Overlap measures how many domains, topics, and curiosities you share. Skill Complementarity checks whether your abilities fill gaps in each other's needs. Problem Alignment identifies if you are working on similar challenges. Communication Style assesses energy, depth preference, and conversational bandwidth. Transmission Match evaluates how well your current broadcasting state aligns with theirs. Synastry, if both users opt in, adds an astrological compatibility layer for those who value it.
Beyond surface-level interests, Serendipity builds a behavioral personality model that deepens over time. The Big Five personality traits (openness, conscientiousness, extraversion, agreeableness, neuroticism) are inferred from your behavioral patterns within the app: how you respond to encounters, when you are active, how you craft your transmissions, and what types of connections you sustain.
Schwartz human values are inferred from your exchange orientation, the problems you prioritize, and the glyphs you choose to leave for others. Attachment style (secure, anxious, avoidant, or disorganized) is modeled from your approach and avoidance patterns across encounters. Critically, none of this is ever asked directly. There are no personality quizzes. Every signal is inferred from natural behavior so the model reflects who you actually are, not who you think you are.
Explore personality science →Personality tells you who someone is. Cognitive resonance tells you how they think. Serendipity models four thinking-style dimensions: convergent vs. divergent thinking, abstract vs. concrete reasoning, integrative complexity (how many perspectives they naturally hold), and epistemic style (how they form and update beliefs).
These dimensions are inferred from text analysis of your transmissions and behavioral patterns across encounters. Two people with complementary cognitive styles often produce the most generative conversations, so the matching engine does not simply seek similarity. It optimizes for productive friction where it matters and alignment where it helps.
Cognitive science behind matching →Raw compatibility is not enough. A 95% match at 2 AM when you are exhausted is not a good encounter. Serendipity layers contextual intelligence on top of the resonance score to predict actual encounter quality. Factors include time of day and your historical engagement patterns, zone quality (how previous encounters in this location went for you), relationship depth if you have met before, and estimated fatigue level from session activity.
The final prediction blends personality alignment, cognitive compatibility, values resonance, and these real-time contextual signals into a single encounter quality estimate. The system optimizes for encounters you will be glad you had, not just people who look good on paper.
Explore the protocol →Trust is not binary. It develops across multiple encounters, and Serendipity mirrors this reality with five progressive depth layers. Each layer reveals more nuanced personality insights and deeper compatibility analysis as you build genuine rapport with someone over time.
Surface (gray) reveals basic interest overlap and energy compatibility from the very first ping. Context (blue) unlocks after a mutual acknowledgment, showing communication style and skill alignment. Character (purple) emerges after 2-3 encounters, surfacing Big Five personality data and values alignment. Shadow (amber) activates after 5+ encounters, revealing projected vs. perceived self and blind spots reframed as strengths. Resonance History (green) is the deepest layer, showing full encounter history patterns and the evolution of your connection over time.
Progressive depth in the protocol →After five or more encounters with different people, the Shadow Profile activates. This is one of Serendipity's most distinctive features. It compares two images of you: the self you project through your transmissions, behavior patterns, and glyph choices, and the self others experience, as captured through the glyphs they leave for you and the encounter patterns that emerge.
The gap between these two images is where the deepest insights live. Hidden strengths you did not know you had. Blind spots that, reframed constructively, become growth edges. The Shadow Mirror never judges. It illuminates. And because it is built from aggregate patterns, no single encounter or person can distort it.
Shadow psychology in Serendipity →The mesh learns collectively while protecting every individual. Using differential privacy, each device contributes noise-injected data to aggregate models: zone quality scores that help identify great spots for encounters, glyph effectiveness rankings that surface which reputation signals carry the most meaning, and dimension importance rankings that tune the resonance engine over time.
Every data point is injected with calibrated statistical noise before it leaves your device. The aggregate patterns become clear at scale, but no individual contribution can be reverse-engineered or traced back to a specific person. The result is a system that gets smarter for everyone without anyone sacrificing their privacy.
The most sophisticated social matching engine ever built, running entirely on your device.