Read-only by default
Campaign Medic reads the campaign data it needs to find waste. Connecting an account changes nothing.
Ad account diagnostics
Connect your ad accounts, uncover costly problems and approve the fixes, without giving AI uncontrolled access to your budget.
No card required · Read-only until you approve a change
Your account stays yours
Campaign Medic reads the campaign data it needs to find waste. Connecting an account changes nothing.
Each fix is shown as the exact change it will make. Nothing reaches the ad platform until you approve it.
Budget caps, spend ceilings and protected campaigns apply to the dashboard, your AI assistant and API keys alike.
Every applied change is logged and measured. Supported changes undo in one click, and you can disconnect at any time.
How it works
The same three steps whether you work in the dashboard or from your AI assistant.
Sign in with Google, Meta, Pinterest, Reddit or X and choose the ad accounts to include. Search Console and GA4 add context.
One sign-in covers Ads, GSC and GA4Multi-account and MCC supportRead-only until you approve a changeAsk what is wrong, in the app or from Claude, ChatGPT or Gemini. Your assistant works from live campaign data, not stale exports.
“Why did CPA rise?”“Which campaigns waste budget?”“What should I pause?”Each fix is drafted as an exact change. Guardrails check it on the server, you approve it, and supported changes can be undone later.
Approval required for every writeGuardrails on budgets and brand campaignsUndo for supported changesGuardrails
Protect brand campaigns, cap how far a budget can move, or block pausing outright. The rules run on the server before any write reaches the platform, so they hold for the dashboard, Claude and API keys alike.
Example rules
Impact monitor
Campaign Medic records the metric before it touches anything, then measures the same window afterwards. Where the platform supports it, undo a change in one click from the operations log.
Wasted spend in the 14 days before and after each change. Bars drawn to scale within each change.
Also included
MCP connector
One hosted server and one sign-in approval. Add the address to your assistant and work from the conversation you are already in.
Claude · ChatGPT · Gemini · Cursor · any MCP client
I’ll check search terms, budgets and recent changes before recommending anything.
› get_search_terms · last 30 days
› get_keywords · quality and cost
Found 3 issues. I’ve drafted the fixes for you to review.
18 negative keywords · 2 campaign budget cuts · fully reviewable
✓ Guardrail check passed: budget changes are within your +50% limit
Who it is for
Audit every client account from one workspace, with guardrails and reporting built in.
Catch wasted spend between reviews and see what every approved change delivered.
Manage more accounts without adding repetitive manual work.
Pricing
Prices in GBP. 7-day free trial on Growth AI. Cancel anytime.
£0
forever
No card required
Get started with AI-assisted ad management.
£69
a month
Bring your own Claude, OpenAI or Gemini key. You control the model and the token bill.
£99
a month
Zero setup. AI usage is included, no API key needed.
Custom
tailored to your needs
Hands-on support from our team, scoped to your accounts and goals. Contact us for a tailored quote.
The alternative
You can. This is what changes when the AI works inside a system built to protect the budget.
| Area | A chatbot and an export | Campaign Medic |
|---|---|---|
| Data | Exports you paste in, already out of date | Read live from the ad platform every time |
| Changes | Copied back by hand, or the AI gets full write access | Drafted for you, applied only after you approve |
| Limits | Nothing stops a bad budget change | Budget caps, spend ceilings and protected campaigns, enforced on the server |
| Record | No record of what changed, or why | Every applied change logged in full |
| Undo | By hand, if you remember what it was | One click for supported changes |
| Proof | No measurement afterwards | Before and after on every applied change |
Never. Every write, from adding a negative keyword to changing a budget, is shown to you as the exact change before anything is executed. Guardrails check it on the server, and nothing changes until you click Approve. Supported changes can be undone from Operations.
Connect an account, run a free audit, and see the findings before you pay anything.
No card required