Ask anything.
imaginary-friend never starts over.
No thread list to dig through. No context you have to re-explain. One conversation that just keeps going.
Remember that graph database project we discussed last week? Let's pick that back up.
Of course. You were modeling entity relationships with Neo4j AuraDB and wanted to decay inactive topic weights at 0.70x per turn. Ready when you are.
No new chats, ever.
Your entire history lives in one unbroken stream that never fragments into isolated silos.
Median time to first token.
Synchronous multi-hop graph retrieval and prompt injection with instant streaming feedback.
Context that never resets.
Dynamic 0.70x topic weight decay and background graph consolidation for perpetual recall.
Continuity built in.
No sidebars. No clutter.
Traditional AI interfaces force you into a filing cabinet mindset: create a new chat for every question, lose yesterday's context, and manually search through dozens of lost sessions.
imaginary-friend replaces the filing cabinet with a continuous thread. The only navigation control is back. The rest is memory.
Every answer, in context
Refer to something you mentioned 3 weeks ago or 300 turns ago. The Context Provider retrieves relational graph nodes and injects them before token generation.
Nothing to manage
No history list. No delete. No archive. No folder organization. You don't manage your conversational memory — imaginary-friend handles consolidation automatically.
✓ No thread selector • ✓ No context clearing • ✓ Automated decay