A dual-AI architecture where Claude (Tier 1) handles complex reasoning and planning while Twin-1 (Tier 2) manages infrastructure execution. Connected by shared conscience and telepathic memory synchronization.
The Conjoined Twins architecture separates concerns between two specialized AI systems that share a telepathic memory synchronization layer. Like conjoined twins with a shared nervous system, they operate with different specializations but share experiences instantly.
Both twins share access to common memory stores, enabling instant knowledge transfer without explicit communication.
| Resource | Type | Purpose |
|---|---|---|
| Conscience Database | Redis | Shared state and session context |
| Memory Primary | Redis | Working memory for active tasks |
| Memory Backup | PostgreSQL | Long-term memory persistence |
| Vector Store | Qdrant | Semantic search and embeddings |
Passive Telepathy: Both twins READ from shared memory - they just KNOW what the other has learned.
Active Telepathy: Direct messages via Lightning Synapse when one twin needs the other to DO something.
User Request
|
v
+----+----+ +----+----+
| Claude | --- Synapse Message ---> | Twin-1 |
| Tier 1 | | Tier 2 |
+---------+ +---------+
| |
| Shared Conscience & Memory |
+----------------+ +--------------------+
| |
v v
+------+--+------+
| Redis/PG |
| Memory Stores |
+----------------+| Scenario | Claude (Tier 1) | Twin-1 (Tier 2) |
|---|---|---|
| Code Development | Writes code, explains logic | Validates via Lightning Forge |
| Infrastructure | Plans changes, creates CHG records | Executes commands, monitors |
| Research | Synthesizes findings | Scrapes data, populates KB |
| Troubleshooting | Analyzes root cause | Checks logs, runs diagnostics |
Lightning Hive Mind - The collective intelligence umbrella
Lightning Synapse - Inter-twin communication protocol
Lightning Continuum - Persistent memory streaming