javimosch percept-ambient
Ambient intelligence mode — continuous context awareness without explicit commands.
What it does
Runs in the background, building a knowledge graph of conversations, entities, and relationships over time. Your agent passively learns context from ambient speech — who you talk to, what projects are active, what decisions were made — without needing explicit commands.
When to use
- User wants always-on context awareness
- Agent needs background knowledge from daily conversations
- User asks "what do you know about [person/project]?" based on overheard context
Requirements
- percept-listen skill installed and running
- percept-summarize skill installed (for entity extraction)
How it works
- All conversations are continuously captured and summarized
- Entities (people, companies, projects, topics) extracted automatically
- Relationships mapped between entities (works_on, client_of, mentioned_with)
- Context packets assembled on demand for any agent action
- Full-text search (FTS5) + vector search (LanceDB) for retrieval
Context packets
When your agent needs context, Percept assembles a Context Packet:
{
"recent_conversations": [...],
"resolved_entities": [...],
"relationships": [...],
"relevant_history": [...]
}
This gives the agent rich situational awareness without loading entire conversation histories.
Vector search
Semantic search over utterances using NVIDIA NIM embeddings (primary) with all-MiniLM-L6-v2 as offline fallback. Stored in LanceDB (local, zero-infra).
# Search via dashboard (port 8960) or API
curl localhost:8960/api/search?q=project+deadline&mode=hybrid
Privacy controls
- All data stored locally in SQLite + LanceDB
- TTL auto-purge (configurable retention periods)
- No audio stored — only transcripts
- Dashboard → Settings → Privacy for granular controls
Real-time dashboard
Monitor ambient intelligence at http://localhost:8960:
- Live conversation feed
- Entity graph visualization
- Search across all conversations
- Analytics and usage stats