← All posts

June 21, 2026 · 5 min read

Set Up AdWhispr in ChatGPT: Competitor Ads + Campaign Launch in One Connector

Set up AdWhispr in ChatGPT in five minutes. One connector gives you competitor ad research, ad cloning, and live campaign launch — with prompts to try first.

ChatGPT is already where a lot of media buyers draft copy, brainstorm angles, and argue with themselves about budgets. What it can't do out of the box is see what your competitors are actually running — or ship a campaign. Setting up AdWhispr in ChatGPT fixes both in one connector: 27 MCP tools that research competitors' proven winners, clone the best ad for your brand, and launch it as a live campaign without leaving the chat.

This is the ChatGPT companion to our Claude Code setup guide and Cursor setup guide. Same server, different client. About five minutes, start to finish.

What you're connecting to

AdWhispr started as competitor-ad research over chat. As of June 2026 it closes the full loop. The 27 tools break down into four groups:

The research side reads from the Meta Ad Library, TikTok ad research, and Google keyword intel. At the time of writing that's 380+ tracked brands and roughly 194,000 ads — including 74,000+ ads that have run 100 days or longer. Days-running is the performance proxy for everything: brands don't keep paying for losing ads, and we track it from daily snapshots instead of inventing ROAS numbers nobody can verify.

Set up AdWhispr in ChatGPT

ChatGPT supports MCP servers through connectors (App integrations). The exact endpoint and current walkthrough live at adwhispr.com/integrations — that page always matches the latest ChatGPT UI, so treat it as the source of truth. The short version:

  1. Open ChatGPT's connector settings. In ChatGPT, go to Settings → Connectors and choose to add a custom connector.
  2. Paste the AdWhispr MCP endpoint from adwhispr.com/integrations.
  3. Authenticate. You'll go through OAuth with your AdWhispr account on first connect. No account yet? The free tier works for this — more on that below.
  4. Enable the connector in a chat. Start a new conversation, turn the AdWhispr connector on, and ask ChatGPT to list its AdWhispr tools. If you see the research, creative, execution, and management tools, you're live.

If you'd rather run the server locally (or wire it into a terminal client alongside ChatGPT), the npm package is adwhispr-mcp-server — unscoped — and npx adwhispr-mcp-server config emits the config block. That's the route the Claude Code guide walks through in detail.

First prompts to try

Start with research — it's read-only, it's the fastest "wow," and it doesn't touch any ad account.

Add Liquid Death and show me their longest-running active ads.

That fires add_brand (or search_brands if the brand is already tracked), then get_brand_ads sorted by longevity. You get the proven winners, not whatever ran yesterday.

Compare these three competitors' ad strategies and tell me where the gap is.

compare_brands does a side-by-side strategy diff: formats, hooks, offers, iteration rate.

What keywords is [competitor] bidding on that I'm not?

research_competitor_keywords pulls Google keyword intel conversationally — we wrote a full post on competitor keyword research with an AI assistant.

Then, once you've connected an ad account with connect_ad_account, the full-loop prompt:

Find the longest-running ad from @competitor, clone it for my brand,
and launch it on TikTok with a $50/day budget.

Research → clone → launch, one message. If you want eleven more prompts like that, we published a full library: 12 prompts that take you from competitor research to a live campaign.

What the free tier includes

Plan Price What you get
Free $0 5 MCP tool calls/month, 1 brand
Pro $39/mo (3-day free trial) Unlimited research tool calls, 3 brands, 10 ad clones/month
Agency $149/mo 10+ brands, 50 clones/month, cross-brand comparison

Five tool calls a month is deliberately small — it's enough to confirm the connector works and run one real research pass, not enough to live on. A single "add this brand and show me their winners" prompt typically burns two or three calls.

One thing worth noticing about the pricing: it's flat. Pro is unlimited research tool calls, full stop. Some tools in this category meter every tool call and charge overage per call, which turns every question you ask your assistant into a line item. We think that's backwards — you should be encouraged to interrogate the data, not billed for curiosity. If you're weighing the metered alternative, the AdWhispr vs Adspirer breakdown covers where each tool fits.

If you clone heavily, one-time credit packs top you up without a plan change: 10 clones for $19, 30 for $39, 100 for $99.

The upgrade path, honestly

Here's the sequence most people follow:

  1. Free: connect the ChatGPT connector, add one competitor, pull their longest-running ads. Decide if the data is real. (It is — and when it isn't available, we say so. Meta's Ad Library doesn't expose CTR, CPC, or ROAS, so we never print them.)
  2. Pro: unlimited research means you stop rationing questions. Three brands covers you plus your two most annoying competitors. Ten clones a month covers a steady testing cadence.
  3. Agency: you're running 10+ brands or you're an actual agency. Cross-brand comparison becomes the weekly ritual.

The 3-day Pro trial means you can run a genuine research-to-launch loop before paying anything.

Where to go from here

Once the connector is live, ChatGPT stops being a copywriting intern and starts being an analyst with hands. It can tell you which competitor ad has run for 200+ days, clone it in your brand's voice, and put it live on Google Search, Performance Max, or TikTok — then pause it, resume it, or move budget when you ask.

If Claude is your daily driver instead, the same server works there too — see how to run ads from Claude in 2026. Either way, setup is once; the loop is forever.

Start free at adwhispr.com, or grab the endpoint at adwhispr.com/integrations.

One connector, 27 tools, zero tabs.