Personal AI
An assistant that knows its user.
Preferences, people and plans from every conversation, kept as each user's own memory, which they can inspect, correct and export.
What changes
With Geniffy.
An assistant that knows its user. Preferences, people and plans, recalled before the model answers.
Existing history, imported. What other assistants already learned can be imported, so the first day is not a blank start.
Control for every user. Every memory can be inspected, corrected, forgotten or exported by the person it describes.
How it works
For personal ai.
- 01Every conversation in
Each exchange is saved after the reply; only what matters becomes memory.
- 02Known on the next turn
context() puts what is known about the person in front of the model, with sources.
- 03Visible to the user
Each memory can be shown, corrected, forgotten or downloaded.
Three lines, with the Python SDK
mem = client.space(user.id)
mem.memories.add(messages=chat_history) # after each reply
known = mem.context(user_message) # before the next oneFits your tools
Already in your stack.
- OpenAI
- Anthropic
- Gemini
- Telegram
- Google Calendar
Other use cases
More ways to build.
- Customer supportSupport that knows the customer.
- SalesEvery account, remembered.
- Coding agentsDecisions that outlive the session.
- EducationTutors that remember each learner.
- Health and coachingCare that builds on every session.
- Voice agentsEvery call continues the last one.
- CommerceFollow-ups that know the shopper.