Change the AI model under most assistants and you start again. The new model doesn't know how you like to work, what it promised last week, or who it is to you. Every session, and every switch, is a fresh stranger.
Earlier this year, Murage's founder Sean Donahoe built a project to fix that for himself. He called it PIP, short for persistent identity. The idea was simple: what makes an assistant yours should live outside any one model, so it carries across sessions, across memories and across whatever model or system you use that day.
That idea is in Murage today, as Continuity.
What Continuity does
Open a bot's Memory settings and turn on Keep a continuous self. Then write, in your own words:
- Continuity summary: who this bot is to you, in a few lines.
- How we work together: how you like to work with this bot.
- Commitments: what it has committed to.
The bot carries what you write into every conversation, on any engine. Switch it from one AI engine to another and it is still the same bot. Only you can change what it carries.
Optional: reflect between conversations
Turn on Reflect between conversations and, after you talk, the bot reads your own words and suggests things to keep. You choose which conversations it can reflect on. Proposed commitments and traits are added only when you confirm them.
Why you stay in charge
Continuity is a record you write and can read, not something the bot decides about itself. You can open it, change it or switch it off at any time, and What shapes in the bot's settings shows everything it reads before it answers.
A few things to know:
- Continuity needs memory turned on.
- Exporting a bot does not carry its continuity.
Download Murage and give your Chief of Staff a Continuity summary. It's the same Chief of Staff tomorrow, whichever engine it runs on.