Epic: Two Vectors Aware of Each Other
This is post-M8 — the most novel thing in the project.
"Two Vectors in the same room, each running their own VectorBrain instance, each with their own memory and relationship with the humans in the space, but aware of each other."
— MACHINA_ANIMA, Claude entry
What this epic delivers
- mDNS peer discovery between VectorBrain instances on the same network
- Shared semantic memory substrate (shared ChromaDB or observation broadcast)
- Presence signaling: each Vector knows the other is in the room
- Shared person model: Vector A's observations about Dexter available to Vector B
- Non-linguistic inter-robot signaling (orientation, chirp pattern) — not sentences
Why it matters
One Vector knows Dexter from the desk where work happens. One knows him from the living room where he rests. Together they have a more complete picture than either alone.
This has never existed at consumer scale.
Sub-issues
Architecture notes
- Per Claude: interaction should be animal-level social signaling, not linguistic. Head turn + chirp, not sentences.
- Per Gemini: shared ChromaDB instance, two fragmented physical nodes of one episodic memory.
- Per Grok: each Vector has its own memory and personality — the shared layer is observations about humans, not identity.
Definition of done
Vector B says "Did you burn the toast again?" to Dexter walking into the office, based on an observation Vector A made in the kitchen 15 minutes earlier.
Epic: Two Vectors Aware of Each Other
This is post-M8 — the most novel thing in the project.
What this epic delivers
Why it matters
One Vector knows Dexter from the desk where work happens. One knows him from the living room where he rests. Together they have a more complete picture than either alone.
This has never existed at consumer scale.
Sub-issues
Architecture notes
Definition of done
Vector B says "Did you burn the toast again?" to Dexter walking into the office, based on an observation Vector A made in the kitchen 15 minutes earlier.