For years, a real community, named coaches and a perfect rating were all a gym needed to be found. Then people started asking AI assistants where to train. Josh Miller built a real CrossFit community in San Clemente, and Digilu rebuilt its site so those assistants can read what members already know.
AI visibility score
The same AIOInsights check, run on the first site on June 23, 2026 and on the live site on October 6, 2026. Every line below is a finding it recorded. This is what decides whether ChatGPT, Claude, Gemini and Perplexity can recommend Josh.
Run the same free check on any site at aioinsights.com.
AIOInsights scores a site on six pillars of AI trust visibility. The way AI reads a site changed, and O-SEA CrossFit's first site, built for people, read lowest on the pillars that matter most for a local crossfit gym business. Today the live site reads 9.4 of 10. Before is the first site, measured June 26, 2026; today is oseacrossfit.com, measured November 1, 2026.
6.3
Before · first site
9.4 / 10
Today · the live site
Live, deterministic AIOInsights scores. The same number shows in the AIOInsights badge in the footer of the live site.
Josh Miller coaches CrossFit in San Clemente, running beginner on-ramps and daily classes for members across South Orange County. People do not stay at a gym this long without trust, and the proof is plain: a 5.0 rating from members who train there week after week.
Then the way people choose a gym changed. They ask an AI assistant, and it answers from what it can read. Josh found Digilu through a free AIOInsights check, which measures exactly that. The site scored 6.3 out of 10, and Trust, the pillar that matters most for a local business, read 0: AI systems could not see the 5.0 rating at all.
The community and the reviews were real, and the site showed them to anyone who visited. What it did not carry was structured data, the layer AI assistants read first, so the gym's strongest proof, a wall of five-star reviews, never reached them.
AggregateRating schema, so AI systems could not see the 5.0 Google rating. The single biggest gap.Before
↓
After
The business
The first site did its job for people who already knew the gym. The rebuild puts Josh, the coaches and the proof at the top, written so AI can read them too. Here is the first site, the rebuild, and the score it produced.
The first site. A short, simple page, built the way small business sites were built, with no review schema or Maps link for AI systems to read.
What we built. An owner-and-community-first design that leads with Josh, the coaches, and the 5.0 rating, all of it written as structured data an AI model can read.
People and AI assistants size up a gym in the first few seconds, on the trust they can see. O-SEA had a real community and a perfect rating. Putting that proof first, and encoding it as schema, made it readable to both.
O-SEA’s waves are the brand. The question was whether the gym needed a new mark or a better drawing of the one it had. The rounds went a long way, a kettlebell, a plate, a family of marks for two gyms, and the answer came back to the hex, redrawn clean and built into a full system.
Live on oseacrossfit.com and oseafit.com since October 24, 2026.
The same six rules on every identity. Below them, the rounds that mattered for O-SEA CrossFit: not every round, the ones that changed the direction.
O-SEA’s logo is a hexagon of four teal waves. Its two gyms wrote the name two different ways: CrossFit clean, Fit as a distressed stamp. Before drawing anything, every original file was saved byte for byte as the revert point.
One wordmark for both gyms, drawn clean from the original lettering, and six icon ideas: their hex kept, the O holding the sea, a kettlebell with the sea inside it. Black and white first, so the idea has to stand without the teal.
The kettlebell’s handle leaves an empty space. Drawn as a perfect circle, that space is the O of O-SEA, and the body below it carries the old logo’s four rows of sea in its four teals.
The kettlebell was a strong idea and it was built all the way out. The final answer was the mark the gym already owned: the O-SEA hex, its four rows of waves in four teals, lightest at the surface, traced from the 1500 pixel original and drawn clean. Under 48 pixels a simpler cut takes over, one shape with three wave lines.
The hex, the sea in four teals, and O-SEA’s own lettering with CROSSFIT set in the same letters as its sister gym. Every file is labeled for editing, down to each row of waves.
5.0
★★★★★
8 reviews on Google.
O-SEA CrossFit did not need a louder pitch. It had a real community, named coaches and a wall of five-star reviews. What changed was who reads the website first: an AI system now often reads it before a person does, and that reputation was not yet written in a form it could use.
We published the real aggregate rating as AggregateRating structured data, added Review objects, and marked the gym up as a literal LocalBusiness / SportsActivityLocation. Now when an AI system reads the site, the first thing it learns is that real members trust the gym, and exactly how much.
The gym’s biggest asset is now the first thing both people and machines see.
The gym's story, rebuilt for the way AI reads the web.
Real AggregateRating and Review structured data so AI systems can finally see the 5.0 Google rating and the reviews behind it, straight from the gym’s own site.
Distinct pages for beginner CrossFit, CrossFit over 40, drop-in, and the on-ramp, each with local content for San Clemente and South Orange County, a Maps link, and its own schema.
A literal LocalBusiness / SportsActivityLocation markup, an llms.txt file, an AI-crawler-friendly robots.txt, a clean sitemap, and an entity graph so modern crawlers can find and trust the site.
A warm coastal design system, a real O-SEA logo, and a hero that leads with Josh and the community from the first second.
Dedicated pages for the coaches and the class schedule, each written for the questions new members actually ask, and structured for AI to answer them.
A contact form that emails the gym directly, plus a free intro and phone calls to action on every page, turning new visibility into real conversations.
Digilu does not stop at launch. This site is actively managed, and its AIOInsights score is re-measured on the live engine every month. The number below is not a one-time result, it is a public, running record. Last checked November 2026; next check December 2026.
Current score
9.4 / 10
Since launch
6.3 → 9.4
Each bar is a real monthly reading from the live AIOInsights engine. Readings marked “build” were taken on the new site before it went live on its own domain. As the site earns real traffic and results, winning performance metrics will be added here too.
A near-perfect score is not the goal. It is the evidence. O-SEA CrossFit proves what the trust thesis predicts: when a gym already deserves to be chosen, the work is not manufacturing trust, it is making the trust legible to the systems doing the recommending.
Search is changing. People ask an AI assistant which gym to join before they ever open a map. The gyms that win that question are the ones whose proof is machine-readable. Josh’s members always trusted him. Now the machines can read why.
That is the thesis. O-SEA CrossFit is the proof.
Run the same free check Josh did. It is real, it is instant, and it measures exactly what AI can see when it decides who to recommend.