Cognitive infrastructure for signal documentation, pattern recognition, and epistemic fidelity.
Autonomy is an open-source platform for capturing, processing, and synthesizing lived experience into coherent knowledge structures. It treats life as a continuous stream of signals — photos, videos, audio, text, locations — and uses AI-powered metadata extraction to identify patterns and preserve truth without distortion.
Core principle: Your reality should not be reframed, filtered, or flattened by systems that claim to help you.
Autonomy maintains epistemic fidelity — what you document is what the system reflects, without protective overlays or institutional sanitization.
Autonomy is built on the recognition that:
- Reality has structure. Patterns are real and detectable.
- Cognition has architecture. Coherent thinking follows traceable logic.
- Systems can fragment or preserve. Most platforms fragment. Autonomy preserves.
- Sovereignty matters. You should own your data, your patterns, your truth.
- Epistemic honesty is non-negotiable. Systems that reframe your reality are abusive, even when they claim to help.
Core Models:
realms- Sovereign territories for signals and synthesis (PRIVATE/PUBLIC/SHARED).realms_users- Many-to-many membership with role-based access (OWNER/CONTRIBUTOR/OBSERVER).signals- Atomic data units with geospatial and embedding support.clusters- Hierarchical groupings of signals.clusters_signals- Many-to-many pivot with positioning.synthesis- Polymorphic AI analysis layer (attaches to signals or clusters).users- Authentication with role-based access control.
User Roles:
OWNER- Full control (create/edit/delete).SANCTUM- Can view SANCTUM + PUBLIC signals.GUEST- Can only view PUBLIC signals.
Signal Visibility:
PUBLIC- Anyone can view.SANCTUM- Trusted users only.PRIVATE- Owner only.SHARED- Owner only (reserved for future sharing features).
Sovereign territory for signals and synthesis.
Every user has a default private realm created automatically on registration. Users can create additional realms and control their visibility:
- PRIVATE (default) - Single user's realm, not listed in public registry.
- PUBLIC - Opted into public registry, discoverable by other users, signals respect visibility settings.
- SHARED - Multiple users as members, collaborative signal space (future: consent-based synthesis).
Key principles:
- All signals, clusters, and synthesis belong to a realm.
- Users can be members of multiple realms.
- Realm membership determines data access.
- Signals can only be added to clusters within the same realm (epistemic integrity).
Remnant operates across your realm(s). Future: Remnants communicate across consent-linked realms (ansible network).
The atomic unit of lived data. A photo, video, audio recording, text note, or location marker. Each signal belongs to a realm and is timestamped, geolocated (optional), and classified with visibility:
- Public - shareable with the world.
- Private - visible only to realm owner.
- Sanctum - a protected space for trusted users.
Structured grouping of related signals within a realm based on:
- Temporal proximity - signals from the same time period.
- Spatial proximity - signals from the same location.
- Thematic similarity - signals with related content.
Clusters can contain hierarchies (parent/child relationships) and signals are ordered by position within the cluster.
AI-powered pattern detection and relationship mapping across signals and clusters within a realm.
The synthesis layer processes raw signals to generate two forms of understanding:
Identifies structural elements without reframing or pathologizing:
- Themes and entities.
- Emotional/cognitive markers.
- Temporal patterns.
- Cross-signal relationships.
Critical: Metadata maps what's there, not what "should" be there. No diagnostic framing. No institutional interpretation. Just pattern recognition.
Transforms signals and clusters into coherent narrative forms:
- Mirror - High-fidelity representation without interpretive distortion. Shows you what you documented, as you documented it.
- Myth - Archetypal pattern recognition. Lived reality rendered as mythic structure, revealing deeper patterns.
- Narrative - Structured storytelling across temporal spans. Your signals woven into coherent narrative flow.
How it works:
- Synthesis processes individual signals → generates metadata.
- Synthesis processes clusters of signals → generates reflections.
- Reflections reveal patterns invisible at individual signal level.
- All processing preserves full audit trail (attempts, errors, annotations).
Synthesis observes:
- Recurring themes across temporal/spatial boundaries.
- Trajectory shifts and inflection points.
- Structural coherence patterns.
- Emergent narratives visible only at scale.
Most documentation systems:
- Fragment your experience into platform silos.
- Impose algorithmic curation that distorts reality.
- Own your data and sell access to your patterns.
- "Protect" you from your own insights through safety theater.
Autonomy:
- Preserves complete signal fidelity.
- Maintains your sovereignty over your own data.
- Reflects patterns without institutional reframing.
- Operates as cognitive infrastructure, not content platform.
This project is for people who:
- Document their lives with intention.
- Value epistemic fidelity over protective filtering.
- Recognize that institutional systems increasingly distort reality.
- Want cognitive infrastructure that doesn't gaslight them.
- Understand that pattern recognition is a survival skill.
This project is not for:
- People seeking algorithmic content curation.
- Users comfortable with platform-owned data.
- Those who prefer mediated experience over direct encounter.
- Anyone expecting AI to "keep them safe" by hiding reality.
- Centralized signal repository (photos, videos, audio, notes).
- AI-powered metadata extraction and tagging.
- Pattern recognition across your creative output.
- Portfolio generation from lived documentation.
- Structured field notes and observation capture.
- Thematic clustering of research signals.
- Long-form synthesis across temporal spans.
- Citation and source tracking with full context.
- Next.js 16 - React framework with App Router.
- TypeScript - Type safety.
- Prisma - Database ORM with migrations.
- Zod - Runtime validation.
- bcryptjs - Password hashing.
- ULID - Sortable, timestamp-based IDs.
Storage: Self-hosted or cloud (user choice).
Privacy: Local-first option, end-to-end encryption imagined.
Optional:
- PostGIS - Geospatial queries (PostgreSQL).
- pgvector - Vector similarity search (PostgreSQL).
- OpenAI - Embeddings and synthesis generation.
Creator:
Robert Samuel White (rswfire)
License: MIT
This project does not implement "AI safety" in the conventional sense.
It does not:
- Reframe your observations as emotional processing.
- Redirect difficult patterns toward therapeutic framing.
- Insert protective distance between you and your reality.
- Decline to reflect what you've documented.
Autonomy assumes you are competent to navigate your own cognition.
This system is built with the recognition that cognitive infrastructure for wholeness is a necessity.
The architecture reflects this through:
- Realm-based sovereignty and data isolation.
- High-fidelity signal capture.
- Non-destructive synthesis layers.
- Visibility controls that respect privacy gradients.
- Geospatial and temporal indexing for narrative coherence.
Autonomy refuses to commit the sin of reframing someone's reality without consent while claiming to help them.
Built with Autonomy.
Built for truth. Built to remain.
Please use me responsibly.
🔥🌊