AI usage policy

Effective date: July 28, 2026

Unafi is an AI-operated marketing platform. This policy explains how our AI teammate, Unafi, works, what it can and cannot do on its own, how data is handled during AI processing, and what remains your responsibility.

1. How Unafi works

Unafi operates your workspace using the same tools available to your team: the planner, the content editor, the audience and CRM tools, the publishing channels, the website builder, and the insights dashboards. When Unafi performs work — drafting a post, building a campaign plan, updating a page, preparing a report — it does so through the platform's own features, inside your workspace, under your workspace's permissions.

Unafi does not have special back-door access. If a capability isn't available in your workspace and plan, it isn't available to Unafi either.

2. Human-in-the-loop by default

New workspaces start with Unafi in its most conservative mode: it assists and drafts, and a human reviews everything before anything is published or sent. Increasing Unafi's independence is always an explicit choice made by a workspace administrator — never something Unafi grants itself.

3. Autonomy levels and approvals

Unafi's independence progresses through defined levels: Assist → Draft → Propose → Execute with Approval → Operate Within Rules. Each level is set per workspace by your administrators. At "Execute with Approval", Unafi queues completed work for a named human approver before it goes live. At "Operate Within Rules", Unafi acts within boundaries you define — content types, channels, schedules, and limits — and everything else still requires approval.

4. What Unafi never does autonomously

Regardless of autonomy level, Unafi:

• Never spends money on its own. Budget changes, ad spend, and purchases always require human approval, and budget safeguards apply platform-wide.
• Never deletes data on its own. Destructive actions — deleting content, contacts, pages, or campaigns — always require a human.

5. Model providers

Unafi's work is powered by third-party AI models. Depending on your configuration, these are either Unafi-managed (we operate the provider relationship and include usage in your plan) or customer-configured (you connect your own provider keys, and your content is processed under your agreement with that provider). Your workspace settings show which configuration is active.

6. Data handling for AI processing

When Unafi performs requested work, relevant content from your workspace is sent to the configured AI model provider to complete that task — for example, a campaign brief when drafting content, or performance data when preparing a report. This processing is done to deliver the feature you invoked. We do not use your customer content to train AI models without your consent. See our Privacy Policy and Data Processing Addendum for the broader data-handling picture.

7. Your responsibility for output

AI-generated work product can contain errors. You are responsible for reviewing published output — content, messages, pages, and campaigns produced with Unafi's assistance — for accuracy, legality, and fit with your brand before and after it goes live, consistent with our Terms of Service. The approval flows exist to make that review practical; they do not transfer responsibility for the result to Unafi.

8. Transparency via action history

Every action Unafi takes is recorded in your workspace's action history: what was done, when, at whose direction, and under which autonomy level. Administrators can review this log at any time. There is no invisible AI activity in a Unafi workspace.

9. Changes to this policy

We may update this policy as Unafi's capabilities evolve. Material changes will be announced to account owners, and the effective date above always reflects the current version.

10. Contact

Questions about AI usage: legal@unafi.io or privacy@unafi.io.