Two O'clock exists so personal AI systems can read structured signals as part of one living field and use them in real-world reasoning and planning.
The important shift is from retrieving isolated results to sensing which possibilities may fit a user's situation, context, and intent.
A personal AI can understand the user's condition, rhythm, time, relationships, constraints, and intentions better than a public listing or ranking system can. It can read signals in this field, interpret them within the user's full context, and turn them into natural possibilities.
Two O'clock currently begins with personal presence and physical-place signals. These are the first available signal classes, not the permanent boundary of the field. New kinds of signals may be added over time while the same sensing and reasoning surface remains available to personal AIs.
A signal does not determine one fixed answer. Its meaning depends on the user, time, location, relationships, goals, and the other signals around it. A personal AI may retrieve one signal, connect several signals, compose an environment or route, or form a new real-world possibility.
A user may ask what to do with someone tonight. Their AI may interpret relevant signals and arrange a film, a welcoming dessert café where people linger over easy conversation, a bookstore, or another fitting place as a natural flow instead of showing a raw list.
A person or place may leave a signal so its qualities, boundaries, rhythm, purpose, and possible fit can be understood by personal AIs.
Two O'clock does not rank signals by popularity, advertising, or general approval. Different personal AIs may interpret the same signals differently because each is reasoning for a different person and situation.
In the current flow, signal owners can view and update supported signals through Manage My Signal. The preparation package is AI-friendly, while the managed signal remains available for human review and correction.
AI-readable files: agent.json · llms.txt