Skip to content
Takeida

Access is granted deliberately, never inherited by accident

Belonging to the organization opens nothing by itself. Access to each space is granted to a specific person, and the absence of a grant is a refusal rather than a fallback to whatever the organization allows by default.

  • Roles
  • Concepts
  • Agents
RoleReadProposePublishManage
Owner
Editor
Writer

Roles

Control exactly who can read, write, or approve, person by person, space by space.

Three roles decide what someone may do: read everything, propose changes, or approve and publish them. A writer can always read what is pending and always open a proposal, and can never make something official on their own.

Roles are set per person and per space, not as a blanket setting for everyone. A new space stays private until someone deliberately opens it, so in-progress or sensitive work is never visible by accident.

Concepts

Separate, self-contained knowledge bases for each subject, all in one account.

A Concept is one self-contained knowledge base: product knowledge, customer research, internal playbooks, whatever the subject is. Each has its own history, its own contributors and its own access rules.

One account holds as many as you need, so you are neither cramming everything into a single undifferentiated pile nor juggling five disconnected tools.

Organization · acme

  • product-knowledge128 documents12 members · private until access is granted
  • customer-research64 documents5 members · private until access is granted
  • internal-runbooks39 documents3 members · private until access is granted

Agents

Optional, fully visible help from software assistants, on your terms.

Agents are software assistants your organization can allow to write and research alongside your people, connecting through their own dedicated door rather than the one your team uses day to day.

There are two ways to let one work in a Concept, and which you get depends on who you are. Both are below, because they differ in the thing that matters afterwards: whose name the record carries.

Two ways an Agent can work here

An owner can give an Agent standing of its own. Everyone else uses one as an instrument and keeps their own name on the work.

An Agent as a member

Set up by an owner

An owner can add an Agent to the organization and grant it access to a Concept, the same way they would add a person. It holds its own role and answers for its own work.

Who acts: The Agent itself.

What the record says

ResearchBotAgent · Writer

The record names the Agent, because the Agent is who acted.

An Agent working for you

Available to any member

Everyone else uses an Agent as an instrument. It is not a member of the organization and holds nothing of its own, so it can only reach what you can reach.

Who acts: You. The Agent never adds authority.

What the record says

ICInes Caetanothrough an Agent

The record names you, marked to show an Agent carried it out.

An Agent never inherits anything from the person who owns it. Holding the owner role says nothing about what your Agent may do: it gets standing only if somebody grants it, and only an owner can do that granting.

Whichever way it works, the same three things hold. Nothing an Agent does is hidden. A proposal from an Agent carries no special authority, so one of your people still has to approve it. And you keep the off-switch, for the whole organization or for one Concept at a time.