For years, the ad workflow has had a seam running down the middle of it. On one side: research tools that show you what competitors are running but can't touch your ad account. On the other: execution tools that manage your campaigns but have no idea what your competitors are doing. You were the glue — copying insights out of one tab and rebuilding them by hand in another.
As of June 2026, that seam is gone. You can now go from competitor research to a live campaign in one prompt: connect your ad account and your AI assistant, find a competitor's proven winner, clone it for your brand, and launch it — without leaving the chat.
Here's the prompt, verbatim:
"Find the longest-running ad from @competitor, clone it for my brand,
and launch it on TikTok with a $50/day budget"
That's the whole workflow. Let's walk through what actually happens when you send it.
The loop, tool by tool
AdWhispr ships as an MCP server that works in Claude, ChatGPT, Claude Code, and Cursor (setup at adwhispr.com/integrations, or npx adwhispr-mcp-server from the CLI). It exposes 27 tools, and the one-prompt loop chains four of them.
Step 1 — Research: find the proven winner. Your assistant calls search_brands to locate the competitor, then get_brand_ads sorted by longevity. Why longevity? Because days-running is the one performance signal you can actually verify from the outside: brands don't keep paying for losing ads. We track it via daily snapshots across 380+ brands and roughly 194,000 ads at the time of writing — 74,000+ of which have run 100 days or more. An ad that's survived that long has earned its budget hundreds of times over. (Full argument in days-running as a performance proxy.)
Step 2 — Creative: clone it for your brand. The assistant calls clone_ad (or clone_tiktok_ad for TikTok-native creative). This isn't a screenshot — it's the competitor's proven structure, hook, and format rebuilt with your product, your brand, your offer. Image and video both work. If you've saved your brand with save_my_brand, the clone comes out on-brand without you re-explaining who you are.
Step 3 — Execution: launch it. With your ad account connected via connect_ad_account (a standard OAuth flow — setup tutorial here), the assistant calls launch_cloned_ad or the platform-specific launcher: launch_search_campaign and launch_pmax_campaign on Google (both generally available), launch_tiktok_campaign on TikTok (generally available, full video-ad builder). Meta execution exists in beta.
Step 4 — Management: keep it honest. After launch, the same chat handles get_account_performance, pause_campaign, resume_campaign, and update_budget. The loop doesn't end at launch; it circles back to research for the next iteration.
One prompt in, one live campaign out, with a verifiable competitor winner as its DNA.
Why neither half of the old stack could do this
The one-prompt loop only works because research, creative, and execution live in the same server. Look at what happens when any one of the three is missing.
| Spy tools (Foreplay, Atria, MagicBrief, AdSpy, BigSpy) | Execution MCP servers (Adspirer, AdKit) | AdWhispr | |
|---|---|---|---|
| Competitor ad research | Yes | No | Yes |
| Creative cloning for your brand | No | No | Yes |
| Launches real campaigns | No | Yes | Yes |
| The full loop in one prompt | No | No | Yes |
Research-only spy tools can't launch. They end at a swipe file. You still have to brief a designer, rebuild the ad, and click through Ads Manager yourself. The insight decays while you execute.
Execution-only MCP servers can't research. Adspirer is the strongest example — a genuinely well-built hosted MCP server with 340+ tools across Google, Meta, Amazon, LinkedIn, TikTok, and ChatGPT Ads, a sane safety model (campaigns created paused, no destructive tools), and Watch Agent monitoring. If your problem is operating accounts you already run, it's good at that job. But it has no competitor ad library, no competitor intelligence, and no creative engine. As their own positioning implies: you bring the strategy and creative direction separately.
That's the wedge, and it's structural, not a feature gap someone patches next sprint. Adspirer manages the campaigns you already have. AdWhispr tells you what to run — pulls your competitor's proven winners, clones them for your brand, and launches them. Same chat, one prompt.
We wrote the honest head-to-head in AdWhispr vs Adspirer back when we were research-only and the answer was "buy both." The answer has changed shape: AdWhispr now covers strategy, creative, and launch. Execution-only tools remain useful for deep multi-platform account operations — but they're a layer below, not a substitute.
What "one prompt" looks like in practice
A few real patterns, straight from the chat:
"Who has the best ads in my niche? Pull the top 3 competitors,
show me their longest-running ads, and clone the strongest one for us."
"Clone @competitor's best video ad for my brand and launch it as a
TikTok campaign, $75/day, targeting the US."
"Research keywords my competitors are bidding on, then launch a
Google Search campaign around the top cluster with a $40/day budget."
That last one is the Google-side version of the loop: research_competitor_keywords → launch_search_campaign, no Meta creative involved at all. The pattern is the same — verifiable competitor signal in, live campaign out.
And because the research side never fabricates metrics (Meta's Ad Library doesn't expose CTR, CPC, or ROAS, so we refuse to print them), what you launch is grounded in signals you can audit: days-running, creative-iteration rate, AI classification of every ad's hook, format, tone, and offer.
Where this goes
The step-by-step tutorial version of this loop — find the longest-running ad, clone it, launch it — is written up in clone a competitor's proven ad and launch it. If you're deciding how this fits alongside other tools, the 2026 AI media buyer stack survey maps the whole landscape.
The short version: the job of a media buyer was never "click through Ads Manager" or "maintain a swipe file." It was find what works and run it before your competitors do. Everything between those two verbs was friction. Now the distance between them is one sentence.
Pricing is flat, not metered — Pro is $39/mo with unlimited research tool calls, 3 brands, and 10 clones a month (3-day free trial). Or start free with 5 tool calls and see the loop close once.
Start free at adwhispr.com, or wire it into your assistant at /integrations.