You can launch TikTok ads from Claude now. Or from ChatGPT, Claude Code, or Cursor — anywhere AdWhispr's MCP server plugs in. TikTok execution is generally available with a full video-ad builder, and it's wired into the same server that does TikTok competitor-ad research and creative cloning. Which means the complete workflow — find what's winning on TikTok, clone it for your brand, launch it as a live campaign — is now a single conversation.
This is the step-by-step tutorial. Four tools, four steps, real prompts throughout.
Step 1: Connect your TikTok ad account
First, wire AdWhispr into your assistant — setup for Claude, ChatGPT, Claude Code, and Cursor lives at adwhispr.com/integrations. Then, in the chat:
"Connect my TikTok ad account"
That runs connect_ad_account, which takes you through OAuth via TikTok Business Center. Approve it once and you're done — list_ad_accounts will show what you can launch into from then on. No API keys, no developer application, no token babysitting.
Step 2: Research what's actually working on TikTok
Here's where launching TikTok ads from Claude gets interesting, and where AdWhispr splits from execution-only tools. Before you spend a dollar, ask what's already winning:
"Research TikTok ads in the skincare space. What are the
top-performing competitor ads right now, and what do the
winners have in common?"
That's research_tiktok_ads — TikTok competitor-ad research, in the same session that will launch your campaign. Dig into a specific rival:
"Show me [competitor]'s TikTok ads. Which ones have they
kept running the longest, and what's the hook on each?"
Longevity is the signal we trust most: brands don't keep paying to run losing ads, so sustained runtime is the closest thing to verified performance you can get from the outside. We've written a full breakdown of why days-running beats fabricated metrics — and why we refuse to print made-up ROAS scores the way legacy spy tools do. If we can't verify a number, you don't see it.
The research step is what execution-only MCP servers structurally can't do. They can launch whatever you tell them to launch — but they have no idea what your competitors are running, so the strategy has to come from somewhere else. With AdWhispr, it comes from three prompts earlier in the same thread.
Step 3: Clone the winner for your brand
Found the ad you wish you'd made? Don't screenshot it into a brief. Clone it:
"Clone that ad for my brand. Keep the hook structure and
pacing, but swap in my product and my brand voice."
That's clone_tiktok_ad. It rebuilds the winning ad's structure — hook, format, tone — around your product, so you're launching a proven pattern instead of a guess. (If you're new to the ethics and mechanics of this, our guide to cloning competitor ads covers the philosophy: steal the structure, never the brand.) Everything you produce lands in your creative library, retrievable anytime with list_my_creatives.
You can iterate before launch, conversationally:
"Make the first two seconds punchier — lead with the problem,
not the product. And give me a second variant with a
testimonial-style opening."
Step 4: Launch the campaign
Now the part that used to mean thirty minutes in TikTok Ads Manager:
"Launch that as a TikTok campaign. $50/day budget, targeting
US women 25-44, optimize for conversions."
That's launch_tiktok_campaign — generally available, with a full video-ad builder behind it. Your assistant assembles the campaign, shows you the plan, and puts it live on your approval. Budget, targeting, and objective are plain sentences, not form fields.
And because the whole loop lives in one server, the single-prompt version genuinely works:
"Find the longest-running TikTok ad from @competitor, clone it
for my brand, and launch it with a $50/day budget."
Research → clone → launch. One prompt.
Managing the campaign after launch
The same session handles the boring-but-vital follow-through:
| Prompt | Tool |
|---|---|
| "How's the TikTok campaign performing?" | get_account_performance |
| "List my active campaigns" | list_campaigns |
| "Raise the budget to $80/day" | update_budget |
| "Pause it while we restock" | pause_campaign |
| "We're restocked — resume it" | resume_campaign |
Check-in Mondays, adjust budgets Wednesdays, never open Ads Manager. If you also run Google, the same connected setup launches Search campaigns and Performance Max — both GA — so a cross-platform week can stay in one thread.
What this replaces
Count the tools in the old TikTok workflow: a spy tool for research (that can't launch anything), a creative team or editor for production, TikTok Ads Manager for launch, and a reporting dashboard for management. Four surfaces, and the strategy degrades a little at every handoff.
The chat-native version is one MCP server with the full loop: research_tiktok_ads → clone_tiktok_ad → launch_tiktok_campaign → update_budget. TikTok's MCP ecosystem is brand new — we've written a full landscape piece on TikTok Ads MCP options — and AdWhispr is, as far as we know, the only server where TikTok competitor research and TikTok execution live together.
Pricing, plainly
Free gets you 5 MCP tool calls a month and 1 tracked brand — enough to connect your account and run your first research pass. Pro is $39/mo (3-day free trial): unlimited research tool calls, 3 brands, 10 ad clones a month. Agency is $149/mo for 10+ brands and 50 clones. Need more clones? One-time packs start at 10 for $19. Flat pricing — no per-tool-call meter running while you iterate.
Start at adwhispr.com, or go straight to the integrations page to wire it into Claude. Your first TikTok campaign by chat is about four prompts away.