The Agent works on top of everything else in AdStellar. Anything it creates (images, videos, campaigns) shows up in your projects, media library, and campaigns just like work you do by hand — and you can keep editing it there.
What you can do
In one conversation, the Agent can:- Create ad creative — image ads, talking-actor videos, UGC-style videos, and ads built from a landing-page URL or a competitor’s ad.
- Work with AI actors — create custom actors, browse the stock library, and put your product in an actor’s hands.
- Analyze performance — rank your top creatives, copy, and audiences; spot underperformers and trends; compare campaigns; and explain why a metric changed.
- Launch & manage Meta campaigns — build and publish campaigns, clone past winners, build a “greatest hits” of your top creatives, and pause, resume, archive, or adjust budgets.
- Automate (coming soon) — set up rules that monitor and act on your ads between sessions (pause underperformers, scale winners, catch creative fatigue, run daily health checks).
- Answer questions — how-to help, strategy advice, and quick lookups about your account.
How it works
Just talk to it
Describe what you want in everyday language — for example, “Make me a UGC video for my new serum” or “Which of my ads should I cut?” The Agent asks the minimum questions it needs, recommends a sensible default when a request could go several ways, and then gets to work.It knows your account
The Agent can look up your own AdStellar data — products, brand kit, projects, assets, AI actors, campaigns, and live Meta performance — so its answers and creative are specific to you, not generic. It only ever has access to your own team and workspace data.Performance analysis, launching, and campaign management require a connected Meta ad account. If you haven’t connected Meta yet, see Connect Meta. Creative generation works without it.
Nothing happens without your confirmation
The Agent is confirmation-first. Before it generates creative, launches a campaign, changes a budget, or pauses or resumes a campaign, it shows you a confirmation card with the details and waits for you to click Confirm. The exact credit cost is always shown up front, so there are no surprises. Click Cancel and nothing happens.Long jobs run in the background
Video generation and other heavy tasks run in the background. You’ll see a progress card in the chat, and the finished result drops back into the conversation when it’s ready. You can keep chatting, switch to another conversation, or leave the page and come back.Attach images
You can paste or upload images directly into the chat — for example, a competitor’s ad you want to recreate, or a product photo you want featured. The Agent can analyze them and use them as references.Conversations
Start a new chat
Open the Agent and type your request, or pick a starter from the launcher icons on the home screen. A new conversation is created as soon as you send your first message.
Pick up where you left off
Past conversations are saved in the left sidebar. Click any one to continue it — the Agent keeps the full context of that thread.
Modes & credits at a glance
The Agent has three modes that trade speed for depth — Quick, Standard (the default), and Deep research. You pick the mode in the composer. Messages use AI credits based on the work done: simple replies and quick follow-ups are free, and heavier requests cost more, up to a per-message maximum that depends on the mode. The exact cost is shown on every reply. Learn more in Modes & Credits.Quick start
Open the Agent
Log in to AdStellar — you’ll land on the Agent home screen. (You can also click Agent in the sidebar.)
Ask for something
Try a starter prompt or type your own, e.g. “Make an image ad for my best-selling product” or “Show me my top creatives from the last 30 days.”
Tips & FAQ
Prompt ideas, best practices, and answers to common questions about the Agent.