Picasi Feature
The role model: who may do what – people and AI alike
Picasi has three roles: read, write, and admin. With them you involve different users appropriately – including external partners. The decisive part: AI access through the MCP integration is governed by the same role model. An AI tool can never do more than the role behind it allows.
Which problem does the role model solve?
Market monitoring is teamwork, but not everyone involved needs the same rights. Management wants to read along, not configure. The agency should maintain sources but not manage users. Without graded rights this ends either in "everyone may do everything" or in one person becoming the bottleneck.
Since AI tools started accessing company data directly, a second question arrives that many vendors leave open: what exactly is the AI allowed to do? Picasi answers it by routing both paths through the same door.
What does the role model do, concretely?
- Three roles – read, write, admin – are graded enough for real division of labor and simple enough to be understood without a manual.
- AI access through the MCP integration is subject to the same permissions as the app – no second, separately maintained permission system for AI tools.
- External partners can be involved – agencies, consultants, and freelancers get exactly the access their task requires.
- Access is granted per workspace – combined with Multi-Workspace, everyone sees only the environment they were granted.
- The number of users is unlimited in every plan – billing counts updates, not heads. Reading along costs nothing extra.
What does the role model deliberately not do?
It is not a fine-grained permission system. Three roles are a deliberate decision against complexity: permissions nobody can survey any more are less safe in practice than a few clearly understood ones. If you need field-level rights, Picasi is the wrong tool.
And it does not replace care when connecting AI. That an AI can only act within its role is a hard boundary – but which role you give an AI connection remains your decision. When in doubt, the smaller one: read access is enough for most evaluation cases.
What does this look like in practice?
Management reads along
Read access is enough: view reports and updates without anyone accidentally changing the source configuration.
The agency maintains the sources
Write rights in the assigned workspace – the agency works independently, without access to other clients or to user management.
The AI assistant gets read rights
Read access is enough for evaluations in your own chat. The role bounds what the tool can do – regardless of what anyone asks it.
Frequently asked questions about the role model
Permissions you'll still understand in a year
Try Picasi with 300 free updates – unlimited users in every plan.