Substy's AI memory system gives the AI the ability to remember details about each fan across multiple conversations. Instead of starting fresh every time a fan sends a message, the AI recalls key information — like the fan's name, interests, preferences, and conversation history — to deliver more personalized, natural responses.
This long-term memory is powered by vector storage, a technology that stores conversation summaries in a way that lets the AI quickly retrieve the most relevant details when responding to a fan.
Memory is generated automatically at specific trigger points during conversations. You do not need to manually enter or manage memory — the system handles everything behind the scenes.
The AI creates or updates memory at three key moments:
1. First Conversation — When the AI engages with a fan for the first time, it begins building a memory profile.
2. Every N Messages (Chunk Size) — After a set number of messages, the AI processes the recent exchange and stores a summary.
3. Phase Reset to Discovery — When a conversation resets to the discovery phase, the AI updates the memory with everything learned in the previous session.
Fan Preferences — What the fan likes, dislikes, and responds positively to.
Conversation History — Summaries of past interactions, including what was discussed and how the fan responded.
Relationship Context — The stage of the relationship, rapport level, and emotional dynamics.
Personal Details — Any personal information the fan shares, such as their name, location, occupation, or hobbies.
The AI can greet returning fans by name and reference previous conversations naturally.
The AI avoids repeating questions it has already asked.
The AI can reference shared experiences from past conversations, creating a sense of continuity.
The AI adapts its approach based on what has worked before.
The AI maintains a consistent personality across sessions.
If you enable AI on a creator account that already has conversation history, the memory system supports backfill. This means the AI can process historical messages to build memory profiles for fans you have already been talking to.
Everything happens in the background as conversations progress. You do not need to enter fan details manually, tag conversations, review memory entries, or trigger memory updates manually. The AI continuously learns and refines its understanding of each fan.
More Personalized Responses — Fans feel like they are talking to someone who genuinely knows and remembers them.
Consistent Personality Across Sessions — Memory ensures the AI maintains the same tone and awareness of the relationship.
Better PPV Targeting — By remembering fan preferences, the AI can make smarter decisions about which content to send.
Scalable Personalization — Deliver a personalized experience at scale without your team needing to remember details about every fan.