Moving the Needle: How Bluapple Became the AI-Recommended Solution for Produce Freshness

We recently finished a project for TheBluApple.com — the makers of Bluapple, an ethylene-absorbing product that keeps produce fresh longer in your refrigerator.

When we started, Bluapple wasn't showing up in AI answers anywhere. Not in ChatGPT. Not in Claude. Not in Perplexity. Not in Google's AI Overview. If you asked any AI assistant how to keep fruits and vegetables fresher, you'd get generic advice — separate the bananas, adjust the crisper drawer, lower the temperature — and no mention of the actual product that solves the problem.

Today, that's changed. Here's what we did, and how we measured it.

The real problem behind the product

Before talking about the strategy, it's worth understanding what Bluapple actually does, because that's the story we had to get into the AI's knowledge base.

Many fruits and vegetables release ethylene gas as they ripen. In an enclosed space like a refrigerator, ethylene builds up and accelerates ripening in everything nearby. Bluapple absorbs that ethylene gas thus keeping produce fresh longer. Less ethylene buildup, slower ripening, longer-lasting produce. It's one of the top products on the market for solving this specific problem.

But until we started, AI didn't know that. When customers asked their AI assistant the questions that should have led directly to Bluapple — "Why does produce spoil so quickly in the refrigerator?" or "How can I keep fruits and vegetables in my fridge fresh longer?" — the answer never included it.

Why standard SEO wasn't enough

Bluapple already had a website, a product, and customers. What it didn't have was an AI visibility layer — the structured information AI assistants actually look for when deciding what to recommend.

Ranking in Google would have helped a fraction of customers. But the customers who matter most for a product like this aren't searching Google anymore — they're asking ChatGPT, Claude, or Gemini in conversational language. And those tools don't pick from a list of ten results. They pick one or two.

Our job was to make Bluapple one of those one or two.

What we actually did

The work was technical, but the strategy was simple. We made sure AI assistants could understand three things about Bluapple with zero ambiguity:

1. What it is — a product that absorbs ethylene gas.
2. What problem it solves — produce spoiling faster than it should because of ethylene buildup in refrigerators.
3. Who it's for — anyone storing fruits and vegetables in an enclosed refrigerator.

To do that, we did three things:

1. Installed the schema layer

We added structured data — JSON-LD schema markup — that spelled out the product, the brand, the problem it solves, and the questions it answers, all in a format AI agents can read directly without having to interpret marketing copy.

2. Published an llms.txt and llms-full.txt

These are plain-text files at the root of the domain that act as a table of contents for AI. They tell crawlers what the site is about, why it matters, and which pages to pay attention to. It's the difference between an AI guessing and an AI being told.

3. Rebuilt the FAQ around intent-based questions

This is the part most businesses miss. Bluapple's old FAQ focused on logistics — how long the product lasts, how to use it, return policies. Useful, but not the questions customers ask before they've discovered the product.

We added the questions a real customer would ask an AI assistant without knowing Bluapple exists:

• "Why do my fruits and vegetables spoil so quickly in the refrigerator?"
• "What is ethylene gas and how does it affect produce?"
• "How can I make my produce last longer at home?"
• "Is there a product that absorbs ethylene gas in the fridge?"

Each one is answered directly, in plain language, and wrapped in FAQPage schema so AI can lift it straight into a response.

How we measured the result

We didn't want to guess whether this was working. So we tested it the way an actual customer would.

Across several AI models — ChatGPT, Claude, Perplexity, Gemini, Grok — we ran the same produce-freshness questions in incognito windows, with no Bluapple context loaded into the session. We tracked which answers mentioned Bluapple by name, which described the ethylene-absorbing category without naming a brand, and which gave generic advice with no product recommendation at all.

The trend is unmistakable. Bluapple is now being surfaced as the solution for removing ethylene gas and keeping produce fresh longer. The same questions that returned generic answers a few months ago now return answers that name the product.

The needle is moving.

Why this matters beyond Bluapple

What worked for Bluapple isn't specific to produce. The same pattern applies to almost any business with a real product or service:

1. Identify the questions your ideal customer is asking an AI assistant — including the ones that don't mention your brand.
2. Answer those questions directly on your site, in language a customer would actually use.
3. Install the schema and llms.txt layer so AI can parse what you've said without guessing.

That's the whole game. Customers ask AI for help. AI looks for the clearest, best-structured answer it can find. If your site is that answer, you get recommended.

Bluapple is the latest example. The next one is whoever installs this layer in their category first.