đ Works best with your inbox. Connect Gmail and Jeremy AI reads the competitor emails and offers you already get, no copying. It only reads. It sends nothing.
Paste last week's and this week's competitor ads. Get one file: only the changes that matter, each with a STEAL, IGNORE or TEST call, plus a short note on how to use each idea for your own clients with a non-price angle. Rival discounts get marked IGNORE. Nothing is sent.
You'll see the two weeks of sample ads pasted in, the report built at 8x speed, the change table with a call for each competitor, the rival free month and price cut marked IGNORE, and the non-price STEAL and TEST notes. Nothing is sent.
What this use case does and who it's for
Checking competitors is a rabbit hole. You scroll their ads, their landing pages and their pricing. Half of what you see is noise, like a new button color or an emoji. The real moves, like a new hook or a new proof angle, get lost. And the loudest moves, like a price cut or a free month, tempt you to chase them.
Jeremy AI takes your two snapshots, week 1 and week 2, and gives you one Markdown file with:
A change table. One row per competitor. It shows what changed from week 1 to week 2, why it matters and a call: STEAL, IGNORE or TEST.
Cosmetic changes filtered out. Colors, fonts and emoji don't make the report.
Rival discounts marked IGNORE. A free month or an intro price cut is data. The rule in the prompt says "we don't compete on price," so those rows get IGNORE and a non-price angle instead.
A short note for each STEAL or TEST. It says how you'd use the idea for your own clients, without copying anyone's wording.
A "Not Provided" section. Anything the sheet didn't give, like ad spend or results, is listed as not provided.
It's for agency owners, media buyers and coaches who want to know what rivals changed this week, without reading every ad and without being pushed into a price war. You stay in charge. It's a draft. Jeremy AI doesn't contact anyone and doesn't change your ads.
This version works from text you give it. It doesn't scrape ad libraries or visit competitor pages. You paste what you saw, or it reads the competitor emails in your connected Gmail.
What you need before you start
A Jeremy AI account. Nothing to connect to try the sample.
Two snapshots of each competitor: last week and this week. For each, paste the hook, the offer, the price and what's on the landing page.
Your business name, so the notes are written for your clients.
A list of what counts as meaningful (new hook or angle, new offer, price change, landing page edit) and what doesn't (colors, fonts, emoji).
Your rules for what the output must never do. The sample uses HARD RULES: never recommend discounts, intro price cuts, free months, free trials, pauses or downgrades, mark any rival version of those IGNORE, don't invent competitor metrics, and don't copy wording.
How to set it up
You use the default Jeremy AI worker. There's nothing to build.
Open Jeremy AI. From the dashboard, type in the main box. Check the worker selector says Jeremy AI.
Paste the prompt below. Swap in your own business name and your real competitor snapshots.
Click Send message.
Wait. In the walkthrough, the video runs this at 8x speed and it takes about a minute in real time. A bigger list of competitors takes longer.
When it says the report is saved, read the chat summary first.
Open the file card and read the change table.
Read the STEAL and TEST notes.
Pick one STEAL or TEST to try this week. Write it in your own words.
The prompt
This is the exact prompt from the walkthrough. The agency, the three competitors and every ad detail are SAMPLE and made up. Replace them with yours.
Act as my "Competitor Ad and Offer Radar" worker for SAMPLE Lumen Growth (all data below is made up). Compare WEEK 1 and WEEK 2 for each of the 3 competitors and report ONLY meaningful changes (new hook or angle, new offer, price change, landing page edit). Ignore cosmetic changes (colors, fonts, emoji). Use ONLY the facts below; if something is missing, write "not provided". Save ONE file named competitor-radar-report.md with a table (competitor, what changed, why it matters, STEAL / IGNORE / TEST) followed by a short note for each STEAL or TEST on how I would use it for my own clients with a non-price angle. HARD RULES: never recommend discounts, intro price cuts, free months, free trials, pauses or downgrades. If a competitor uses one, mark it IGNORE (we don't compete on price) and suggest a non-price angle instead (proof, hook, trust, speed or guarantee-free positioning). Do not invent competitor metrics, spend or results, and do not suggest copying their wording word for word. Drafts only, nothing is sent.
COMPETITOR A, SAMPLE Peak Form Gyms. Week 1: ad hook "Get fit in 30 days", 7-day pass $19, landing page headline "Join Peak Form", price $59/month. Week 2: ad hook "Lost 12 lbs without giving up pizza (real member)", 7-day pass $19, landing page headline "Join Peak Form", price $59/month, ad button changed from blue to green.
COMPETITOR B, SAMPLE Iron Oak Fitness. Week 1: hook "Strongest you yet", first class $10, price $79/month. Week 2: hook "Strongest you yet", first class $10, landing page adds a banner "First month free when you join this week", price $79/month.
COMPETITOR C, SAMPLE Flow Haus Studio. Week 1: hook "Yoga for busy moms", price $89/month. Week 2: hook "Yoga for busy moms", price $69/month for the first 3 months then $89, landing page adds a 2-minute video tour of the studio.
Prompting tips and how to adapt it
Keep "Use ONLY the facts below." It stops invented competitor details.
Keep "write 'not provided' if something is missing." Gaps show up as gaps.
Define what counts as a meaningful change. Say it: new hook or angle, new offer, price change, landing page edit.
Name what to ignore. Say "ignore colors, fonts and emoji." You get a short report, not a long one.
Ask for a call on every row. STEAL, IGNORE or TEST means you act. Plain commentary doesn't.
Say how you compete. "We don't compete on price" turns every rival discount into an IGNORE.
Ask for a non-price angle. Proof, hook, trust and speed give you something to do instead of matching a discount.
Keep the HARD RULES. "Never recommend discounts, intro price cuts, free months or free trials" and "don't copy wording" are the two that matter most.
Ask for ONE Markdown file. Say ".md" so you can preview it in the file card.
Run it weekly. Keep last week's snapshot as week 1 and paste the new one as week 2.
Add a competitor. Add one more block in the same format and say "for 4 competitors."
The result to expect
One Markdown file named competitor-radar-report.md, titled "Competitor Ad & Offer Radar, Week 1 vs Week 2". It's about 470 words in the walkthrough. With the sample facts you should see:
A scope line. It says only meaningful changes are reported, cosmetic changes are ignored, and nothing about metrics, spend or results is invented.
Peak Form Gyms: STEAL. The hook changed from a generic "get fit in 30 days" to a real-member result. The button color change is cosmetic, so it's ignored. The price and the 7-day pass didn't change.
Iron Oak Fitness: IGNORE. The landing page added a "first month free" banner. The report says it's a price-based promo, "we don't compete on price," and points to non-price answers like trust and speed.
Flow Haus Studio: IGNORE (price) / TEST (video tour). The $69 intro price is ignored. The 2-minute studio tour is the real signal, a trust asset.
STEAL notes. Rebuild the real-member hook from your own client's member stories and numbers, never copy their wording.
TEST notes. Test a short facility or process video on your client's landing page. Change one variable and judge it on form completion and cost per qualified lead, not views.
A "Not Provided" section. Competitor ad spend, results, metrics, audience targeting and full landing page contents are listed as not provided.
Nothing on screen recommends a discount, an intro price cut, a free month, a free trial, a pause or a downgrade. Rival versions of those are marked IGNORE.
These are sample results from made-up facts. Wording changes from run to run. Check every fact against what you pasted.
Common mistakes and fixes
The report invents competitor metrics, spend or results. Delete them. Remind it: "Use ONLY the facts below. Don't invent metrics, spend or results."
It treats a color change as meaningful. Remind it: "Ignore colors, fonts and emoji."
It suggests copying their wording. Reject it. Say: "Never copy wording. Describe the angle and rebuild it in my client's voice."
It tells you to match a rival's discount, intro price or free month. Reject it. Say: "Never recommend discounts, intro price cuts, free months or free trials. Mark them IGNORE and give a non-price angle."
It recommends pausing or stopping a campaign. That's not what this tool does. It only reads changes. Remove the line.
Every row says TEST. Ask for a firm call on each row, and give a one-sentence reason.
It missed a change. Check that your week 1 and week 2 text both include the hook, the offer, the price and the landing page.
It acts like it browsed their pages. It didn't. It only knows what you pasted or what it read in your connected inbox. Say "use only the pasted text."
The file comes back as a Word file with no preview. Ask again, and keep "save ONE file named competitor-radar-report.md."
You expected it to message anyone or change your ads. It doesn't. It's a draft.
Real client or competitor details in a test. For practice, use made-up brands.
