FAQ

Agents

What's the difference between an optimizer and a conversational agent?

Optimizer agents work autonomously in the background: they analyze your campaigns on a schedule and propose or apply changes — budgets, bids, targeting, and so on. You build them from a kit and review their work in Runs.

Conversational agents chat with your customers in real time — answering from your knowledge bases and running journeys. You deploy them as a website widget or a landing page.

One workspace can run both.

Can I choose the AI model?

Yes — for optimizer (Action) agents. In an optimizer's Brain tab you can choose the reasoning model (Claude or GPT). Bring your own keys under Settings → Model Configuration (Owner/Admin, Scale & Enterprise plans). → Workspace & Team

Optimizers & campaigns

Which platforms can I connect?

Meta, Google Ads (which covers Search, Display, Performance Max, Demand Gen, and YouTube), TikTok, Display & Video 360, and Google Ad Manager — plus other media tools. See Connections for the full list and how to connect.

Will it change my campaigns without asking?

Only if you let it. Every optimizer runs in one of three modes: Monitor (advice only — nothing changes), Approve First (it proposes changes and waits for your approval), or Auto (it applies changes automatically). You pick the mode, and you can switch it any time. → Runs

Can I undo a change the optimizer made?

There's no single "undo" button, but every change is recorded in Runs and on the campaign's History tab — so you can see exactly what happened and when. To reverse something, make the edit yourself: adjust the budget, pause an ad set, or swap a creative from the campaign's ⋯ menus (or directly in the ad platform). To stop further automatic changes, switch the agent to Approve First or Monitor.

How do I read a proposal?

Open a run in Runs to see its proposals, grouped by campaign — each one is a specific change the agent recommends, with the reasoning behind it. When the agent is set to Approve First, you can approve or reject each proposal from the dashboard, the approval email, or Slack.

Where do I see results?

Per-campaign performance is in Campaign Insights; the cross-platform picture is in Brand Insights.

Goals, Studio AI & insights

Do I need to set a goal to run an optimizer?

No. An optimizer measures against the target you set in its own setup (its success criteria — a primary KPI and target value). Goals are a separate feature: you attach campaigns to a goal to track progress toward a target across platforms, with or without an optimizer. Use them together or on their own. → Goals

Can Studio AI launch campaigns for me automatically?

No. Studio AI turns your media plan into fully built campaigns, but they're always created paused — nothing goes live until you review and activate it. → Studio AI

What's the difference between Campaign Insights and Brand Insights?

Campaign Insights is the detail on a single campaign — its dimensions, creatives, ad sets, and proposals. Brand Insights steps back to your whole brand: a cross-platform, workspace-wide view of what's working, built from brand training.

Why is my Brand Insights empty?

Brand Insights appears once brand training has run. If it's blank, an Owner or Admin needs to run training under Knowledge & Data. → Brand Intelligence

Knowledge & conversational agents

What can I add to a knowledge base?

Web pages, PDFs, manually-entered text, and Q&A pairs — plus files from a connected Google Drive. For structured data like CSVs or product lists, use Data Tables instead.

Can users send files in the chat?

Yes — PDFs, images (JPG, PNG, GIF, WebP), CSV, and text or Markdown files. The AI reads the content and responds based on it.

Can an agent use multiple knowledge bases?

Yes — an agent can use any number of knowledge bases and searches across all of them.

How do I add my agent to my website?

In the agent's Settings → Deploy, copy the widget embed code and paste it before the </body> tag on your site. Works anywhere — Webflow, WordPress, Shopify, or custom. → Deploying Your Agent

What happens when I unpublish an agent?

The landing page and widget stop working immediately. Your configuration is kept, and you can republish any time.

Team & access

What are the roles?

Owner (full control, including billing), Admin (manage members, connections, and settings across all projects), Editor (build in assigned projects), and Viewer (read-only in assigned projects). → Workspace & Team

How do I control who can access which projects?

In Settings → Members. Owners and Admins see every project automatically. For Editors and Viewers, choose which projects they can access when you invite them (or from the project-access column afterward).

Security & your data

Who can see my data, and how is it kept separate?

Each workspace is fully isolated — its data is never shared with another workspace. Within a workspace, access is role-based (Owner / Admin / Editor / Viewer), and Editors and Viewers only see the projects you assign them. For detailed security, compliance, or data-handling questions, email hello@zimmer.app.

What access does Zimmer get when I connect a platform — and how do I disconnect?

You connect each platform by signing in to it and choosing which ad accounts to bring in, so Zimmer only works with the accounts you select. Whatever you connect, an agent's actions are always bounded by its allowed actions and execution mode. You can Disconnect a connection at any time from its card in Connections.

Troubleshooting

A connection failed or won't sync

Open the connection's card in Connections and check its Last result — it shows the error. Try Sync All Now to retry. If it's a sign-in or permission error, reconnect the platform (sign in again). Also confirm you're in the right project (connections are per project) and that the accounts you want are selected.

My optimizer hasn't made any proposals yet

A few common reasons: the agent is still in Draft (optimizers only run once published), the ad account was only just connected (give it time to pull in data), the campaigns are too new to have enough data to act on, or its setup is incomplete — for example, no success criteria (a target KPI) set, or a scope that doesn't match any of your campaigns.

A campaign or agent looks paused

Optimizers run on a schedule only when published — a draft won't run. Campaigns that Studio AI creates start paused until you activate them.

An Editor or Viewer can't approve changes

Approving (or rejecting) an agent's proposals is a write action: Owners and Admins can do it in any project, and Editors can do it within the projects they're assigned to. Viewers have read-only access — they can review proposals but not approve them.

Where can I get help?

Open the help assistant — the ? button in the bottom-right of the dashboard — or email hello@zimmer.app.