wenzel orlandmost AI platforms work like this: you put your data in. you train on their infrastructure. you use their models. the AI gets smarter. then you want to leave. and you discover that your 'AI' is actually their AI, trained on your data, running on their servers, accessible only through their API.
this is the SaaS trap applied to AI. and with AI, the stakes are higher. the data you put in is expertise. institutional knowledge. competitive advantage. the stuff that makes your company actually work.
at doobls, ownership means something specific. memory belongs to your workspace and runs on eu infrastructure. you can export your private chats, stories, and episodes from each workspace. shared and published work stays with the workspace; the personal export does not include agent configuration or procedures.
if you leave doobls, you can take that private work with you. agent configuration, procedures, and shared workspace knowledge remain outside the personal export.
it's the business model. when people know they can leave, they stay longer. when they know their data is actually theirs, they put more valuable data in. trust creates depth. depth creates lock-in that's earned.
the AI systems that win will be the ones that respect ownership. people whose judgment is the work won't put their most valuable knowledge into systems they don't control. and they shouldn't.