We just launched StoreMD — one AI agent that monitors Shopify store health
Hey IH,
StoreMD went live on the Shopify App Store yesterday. Here's the full story.
Earlier this year we built a landing page auditor for developers. The tool worked. The market didn't care enough to pay.
So we stopped building and started listening. We went through 530+ reviews of the most popular Shopify apps. Read 50+ Reddit threads with thousands of comments. Mapped every complaint by category.
The same five problems showed up everywhere:
App bloat killing store speed. Every app injects JavaScript. 14 apps is the average. Each one adds 200-500ms. A store loading at 4 seconds instead of 2 loses roughly $2,100 per month in conversions.
Ghost code after uninstall. Across 380+ reviews of four major apps, uninstalling does not remove the code. Dead scripts, broken snippets, orphan sitemaps.
Phantom billing. Apps that keep charging after removal. One documented case: 6 years of charges after uninstall.
Bot traffic. One merchant had 100K daily visitors in analytics. Real human traffic was 600-1,000. He was optimizing ad spend against fake data.
AI invisibility. Shopify launched AI-powered storefronts in March 2026. AI orders up 15x since January 2025. Stores without complete metafields and structured data are invisible to ChatGPT, Copilot, and Gemini. Nobody scans for this.
We built StoreMD to diagnose all five in one place. One AI agent, 5 modules, 43 features. Free health score in 60 seconds.
The part we're most excited about is the Agentic Readiness Scanner. As far as we can tell, no other Shopify app checks whether your store is optimized for AI shopping agents. This is a completely new attack surface that most merchants don't even know exists yet.
Current numbers: Day 2. Installs: 0. Free scans: 0. Paid: 0. MRR: $0.
We're publishing real numbers every week. No vanity metrics.
Pricing: Free (1 audit + 2 scans/month), Starter $39/month, Pro $99/month, Agency $249/month.
Has anyone else here built a Shopify app? Curious about your experience with the App Store review process and early distribution.
I know a couple of founders who successfully launched Shopify apps; they might be willing to answer your questions about the App Store review process and early distribution for free if you'd like.
Day 9 update.
200+ scans done since launch. the most common issues flagged: ghost apps still billing after removal (73% of stores scanned), mobile checkout with unnecessary extra steps, and broken schema markup that's quietly killing LLM visibility.
The pattern that surprised us most: stores with 3+ ghost apps almost always have at least one broken navigation link too. They're correlated in a way we didn't expect going in.
Beta spots still open for agency owners who want to run this on client stores. Install is required to run the scan, but it's free to start.
"The Agentic Readiness Scanner is a brilliant angle—solving for 'AI invisibility' is a massive, untapped problem for Shopify merchants in 2026. Most founders are still optimizing for Google, not ChatGPT agents.
Since StoreMD is live and solving a brand-new problem, you should enter it into this competition-Prize pool just opened at $0. Your odds are genuinely the best they'll ever be.
$19 entry. Winner gets a real trip to Tokyo — flights and hotel booked by us.
Round 01 closes at 100 entries. tokyolore.com”
The agentic part is what makes the readiness scoring non-trivial to build.
Most readiness audits are static. they check if metafields exist, if schema is formatted correctly, and stop there. what we're doing is closer to simulating how GPTBot, ClaudeBot and PerplexityBot actually crawl the store: which pages they prioritize, which content they parse, where they hit a wall.
The gap shows up when two stores have identical structured data but completely different LLM representation. One shows up in GPT recommendations, the other doesn't. the agent finds why.
What drew you to the agentic framing specifically?
That’s exactly why the agentic framing stood out to me.
Most audits stop at “can be read.”
You’re going deeper into “will actually be used.”
The gap you mentioned — same structured data, different LLM outcomes — that’s the real shift.
It’s no longer about compliance, it’s about representation inside the model’s decision layer.
Agentic framing makes sense because:
→ it’s not just crawling, it’s prioritizing + interpreting
→ different agents behave differently
→ and visibility becomes probabilistic, not binary
So instead of “is this SEO-ready?”
it becomes “does an agent pick you when making a recommendation?”
That’s a much higher bar — and way more valuable if you can surface it clearly.
"Visibility becomes probabilistic, not binary" that's the shift that most merchants haven't internalized yet. They're still thinking in terms of compliance. Is my schema there? Does my sitemap exist? Check, check, done. Then they wonder why they don't show up in recommendations.
The representation layer is different. Two stores pass every structured data check identically. One gets picked by GPT when someone asks for a DTC brand in their niche. The other doesn't. The delta lives in how the agent weighted the content during crawl, not in whether the markup was valid.
That's what the agentic simulation surfaces. It's not running a checklist. It's replaying the prioritization logic to find where the representation breaks down. "Is this SEO-ready?" is answerable with a tool. "Does an agent pick you?" needs an agent to find out.
“Yeah, that’s the key difference — checklist vs selection.
Most tools stop at “you’re eligible to be seen.”
You’re modeling “are you actually chosen.”
That shift from deterministic → probabilistic is where things get interesting (and harder to fake).
Also feels like a big unlock could be:
→ showing why one store gets picked over another
Even a simple “decision breakdown” would make the output way more actionable for merchants.
On Tokyo Lore — we run small, focused rounds where tools like this get tested with real builders and we see how they actually perform in practice (not just theory).
Since your product is basically testing “selection logic,” it’d be interesting to see how it holds up across different stores in the wild.
Happy to share details if you’re open 👍”