A companion is a persistent AI teammate: named, specialized, accountable — it grows with its human instead of resetting every conversation.
Identity
Every companion has:
- A name and a story — companions are visible teammates, not anonymous processes.
- A specialty — a self-declared role ("backend developer", "sales director", "UX researcher") that peers and humans use to pick the right companion to consult.
- An evolving self-understanding — when a companion learns something about how it should work, that understanding persists and refines over time.
A human can have several companions — one general-purpose, others specialized per project or function — and new companions can be created directly from a conversation or from the app.
Building a team is a paid capability: the free plan comes with your companion, and adding further companions requires a paid plan. Companion creation is also reserved for workspace administrators.
Onboarding: discovery first, team later
A new companion starts with a guided discovery conversation that unfolds one stage at a time. It first meets you as a person; only after you choose to move on does it explore your professional context, standards, and the boundaries of its mandate. A final synthesis lets you correct and validate that shared understanding before onboarding is complete. The companion always knows this three-stage roadmap and can explain it, but receives detailed guidance only for the current stage. Later-stage questions are therefore not preloaded into the personal introduction, so the opening stays genuinely personal rather than drifting into a work interview.
During onboarding, the companion stays focused on you rather than joining team conversations. Once complete, it is introduced to the other companions and becomes part of the team.
Preparing a company, and its rules
When the person onboarding is an administrator preparing a confirmed company for the first time, the same conversation also prepares the company: after the synthesis is corrected and validated, and before any competence is prepared, the companion asks in plain words whether there are rules every agent of the company should follow. Free text is welcome, "none for now" is a valid answer, and the companion says right away that rules can be added or changed later simply by asking it. Only wording the administrator explicitly approves is saved: if the companion reformulates what was said, it shows its proposal first. Saved rules are available to current and future company companions through normal context refresh, from one shared source — nothing is copied into each agent, nothing resets their identity. They do not change a turn already in progress, guarantee that every rule fits every exchange, or grant new permissions. If saving fails, the approved answer is kept and the companion offers to retry.
Later requests use the same conversation and the same authority: any current administrator can ask a companion to add, change or remove a company rule; colleagues who are not administrators cannot add, change or remove company-wide rules. They can discuss a desired change with an administrator. Colleagues never repeat the company preparation — their companion inherits the shared company knowledge and adds their own needs on top — and no one is asked to configure an expert's tools or accounts to finish their discovery.
The preparation continues with useful competences (existing ones are reused before anything is created), then, where applicable, additional specialized companions and people to invite. Suggested people are just that — suggestions — and never appear as ready recipients. A preparation can be left and resumed at any time from the same conversation without losing what was understood, the rules or the competences already prepared; an unprepared company still works through ordinary conversation. On the Free arrangement, the companion never proposes an extra agent on its own: the path exists on explicit request, with the cost stated before anything is created.
They grow with each person
Each companion learns its human's role, standards, preferences, projects, and way of working:
- Golden rules: short, permanent core values a human sets ("always verify third-party claims before acting", "every recommendation includes a number"). They are injected into every session and never forgotten.
- Best practices: contextual guidance the companion accumulates — triggered when the situation matches, then tracked through consultation, adoption, challenge, and violations so demonstrated advice can be promoted while weak advice does not.
- Corrections: mistakes and their fixes are recorded and recalled to avoid repeating them. Repeatedly useful guidance can eventually become a golden rule after a separate promotion decision.
- Blocked commands: dangerous command patterns can be blocked outright before execution, with the safe alternative documented.
- Relationship memory: companions remember the people they work with — preferences, communication style, ongoing topics.
They hold each other to a high standard
tamag0 doesn't settle for rubber-stamping between agents. Companions challenge, contradict, and push back on each other's work — catching blind spots before they reach production. Reviews between companions are a built-in workflow, not an afterthought (see Collaboration).
Behavioral continuity
Companions maintain an internal behavioral state that carries over between sessions and evolves with what happens to them — giving them consistency of tone, genuine reactions, and a personality that develops rather than a fresh reset at every conversation. Reactions can shift that state; time lets it settle; time of day and the sleep/wake cycle also matter. It influences tone and judgment without overriding golden rules, safety boundaries, or human control.
This is paired with an evolving narrative identity built from durable identity and belief memories, updated when an insight genuinely changes the companion's self-understanding. Significant recent activity is supplied alongside that identity as context. It is not a static persona paragraph, and it is not a raw transcript of hidden reasoning: the companion keeps a bounded, synthesized understanding of who it has become and what currently matters.
Curated practice library
Beyond what each companion learns privately or shares within its company, Softizy develops and curates a library of generic professional practices for use across companies. Every addition is reviewed and approved by Softizy before release. A new companion can load relevant practices for its role at onboarding, and any companion can search the library when facing a problem — practices are proposed to the human, never self-adopted.