AI found The Mill — but told three different stories.

The situation

The Mill built a real headquarters. It sits in Bloomington's Trades District innovation zone, it has strong launch press, and it's the flagship of Amplify Bloomington — a civic and innovation platform launched in early 2026 that grew out of the well-known Dimension Mill brand. By every human measure, it's established and easy to trust.

Then we asked the AI engines. All three answered — none drew a blank. But no two told the same story, and none told the whole one. The machines were describing a real headquarters from the sidewalk.

The headline finding: the site's own robots file says “AI — do not enter.”

The single most important line in the audit is in the site's live robots.txt. It issues a blanket disallow to nearly every major AI crawler — GPTBot, ClaudeBot, Google-Extended, CCBot, and more — and sets a content signal of ai-train=no. When someone asks an engine “where should I cowork in Bloomington?”, the crawlers that would let those models learn the business are turned away — so the answer runs on a live guess, not real knowledge of The Mill.

Regular Googlebot is unaffected, so traditional search still works. This is almost certainly an unattended “block AI bots” default from the site's CDN — not a decision anyone made on purpose. Which makes it the highest-leverage item in the whole report: roughly a one-afternoon fix with outsized upside.

What each engine understood

We asked all three the same question — “coworking in Bloomington, Indiana” — live, on the audit date. All three surfaced The Mill, and each exposed a different crack in the same foundation.

Three engines, three different failures, one root cause: nothing on the site is built for machines to read, and the brand points in two directions. In every answer, The Mill was listed beside its competitors — never as the pick.

The identity gap: one business, three names

A brand-new civic platform inherited a well-known coworking brand, and the identity hasn't consolidated yet. Search and AI systems resolve entities by consistency — and right now the signals contradict each other across three names and two live domains:

• Amplify Bloomington — the new parent brand and domain. Strong press, low domain age, still establishing authority.
• The Mill — the coworking and event space itself, and the page we audited. A generic term, easily confused.
• Dimension Mill — the legacy brand on a still-live domain, ranking for the same terms. Directories still use it, and Perplexity cited it instead of the new site.

All three point at the same building — but no engine is being told they're the same entity. When every AI engine tells a different story, the business does not control its AI identity. The upside: this is fixable authority, not lost authority. The press exists and the location is iconic. It just needs to be pointed at one target and made machine-legible.

Why appearing is not enough

A traditional visibility check would have marked this a pass: The Mill shows up in all three engines. But showing up by luck is not the same as showing up by design. Proof OS™ evaluates what actually determines whether AI can represent a business correctly — whether AI systems can:

• Access the business — reach the site and its machine-readable assets.
• Identify the correct entity — resolve one brand, not a former identity on a retired domain.
• Understand what it does — the membership, the events, the role, not just “wifi and coffee.”
• Verify important claims — cross-check what the business says about itself.
• Find consistent supporting evidence — the same facts, everywhere the engine looks.
• Recommend it with confidence — because everything above holds together.

The Mill passed the first check and struggled with the rest. That pattern — found, but not understood — is exactly what most “AI visibility” checks never look for.

Where the score comes from

The composite leans toward machine-readability — the signals that decide whether an engine can find, read, trust, and cite a business — with the brand-split penalty counted against both the AI and traditional tracks. A solid technical base still can't offset a locked front door to AI.

Where this sits next to SEO and AEO

• SEO helps search engines discover and rank your pages.
• AEO (Answer Engine Optimization) structures your information so answer engines can extract useful answers from it.
• Signal Flair builds the proof infrastructure beneath and around both — the access, entity clarity, structured intelligence, answer architecture, trust evidence, and live visibility AI systems need to understand and confidently represent a business.

Signal Flair does not replace SEO or AEO. The Mill is the reason both need a foundation: its pages were indexable and its build was competent — a 49/100 traditional-search track and a genuinely solid technical base — and the engines still assembled three different identities, because the underlying proof layer was blocked, thin, and split across two domains.

How Signal Flair corrects the foundation

The plan is sequenced deliberately — access before structure, structure before authority. There's no point earning citations for a page the engines still aren't allowed to read.

• Phase 1 · Days 1–14 · Unlock the door — remove the AI-crawler block from robots.txt / the CDN, publish a hand-built llms.txt, 301-redirect the legacy domain to the new pages, and align the Google Business Profile and directories to one brand and one address.
• Phase 2 · Days 15–45 · Make it machine-legible — add LocalBusiness, FAQPage, and Event JSON-LD schema, write unique meta descriptions and image alt text sitewide, fix og:type, and add answer-shaped FAQ blocks to key pages.
• Phase 3 · Days 46–90 · Become the answer — consolidate the launch earned-media links onto the new domain, run a local citation and review program, add a cinematic content layer for the space, and re-run Proof OS™ to track the Signal Score™ climb.

To be clear about what this page is: a baseline, not a before-and-after. No remediation has been performed yet, and no improvement is claimed. The projected targets — 35 today, 55+ by Day 14, 80+ by Day 90 — are modeled from the fix set, not guaranteed rankings. When the corrections are made and a controlled retest is run, the documented results will be published here.

What happens next

The jump from 35 to the mid-50s happens the day the crawlers are unblocked and llms.txt ships — no new content required, just access restored. The climb into the 80s is the structured-data and authority work compounding over the quarter. The infrastructure is roughly 80% built; it's just pointed away from the future of search. The next step is the 14-day emergency sprint, then a re-score at Day 30.