The 160-Millimeter Problem

I. A small, ordinary failure

Here is a transaction that did not happen.

A customer walks into an independent optical boutique on the Upper East Side looking for Kuboraum — the Berlin-designed, Italian-made house whose oversized "mask" frames have become one of the few genuinely architectural statements left in eyewear. He knows what he needs, and he knows it precisely: a wide fit, masculine, roughly 160 millimeters or more across the front. This is not a preference. It is a measurement. Below that number, the frame pinches, the temples splay, and the whole thing sits on his face like a borrowed jacket. The shop has one that fits. One. He likes it, but he wants to see the rest of the range — other shapes in the same construction, other colorways, other lens treatments, the same architecture in a slightly different mood. That's a reasonable request. That's also, in most cases, an additional sale — the customer is asking to be sold more.

The answer he gets is the answer everyone gets: "Call back in September. We'll have a new shipment."

And in September, the shipment lands, and a thousand people call about fifty frames.

This is the entire industry's demand-management strategy, and it is barely a strategy at all. It is a queue with no visibility, no reservation system, no matching logic, and no memory. The customer who wanted exactly what the store will have in six weeks is not on a list. He is not tagged. He is not notified. He is a person who walked out, and the shop's only hope is that he happens to walk back in. Multiply that by the hundreds of independent optical shops across Manhattan and you are looking at one of the most quietly inefficient markets in the city.

II. This is not a service failure. It's an information failure.

It's important to be precise about where the breakdown actually is, because the reflexive diagnosis — retail is dying, go online — is wrong here, and expensively wrong.

The shops in question are excellent. Petite Optique, in the West Village and Union Square since 1995, carries European lines that genuinely cannot be found elsewhere in the city. Occhiali on Lexington has spent three decades building a roster — Barton Perreira, Matsuda, Mykita, Thierry Lasry, Blake Kuwahara, Eyevan, SALT — and a consultative method where the optician asks what you're after and brings out frames suited to the shape of your eyes rather than pointing you at a wall. Euro Optica runs two locations on Columbus and Third with clinical infrastructure behind the retail.

These are not commodity operators. They are curators with deep, specific, hard-won knowledge — exactly the kind of business the internet was supposed to be unable to replace, and hasn't.

The failure is upstream of the store. It's that nothing these shops know is addressable from the outside.

Consider what is actually true inside one of these boutiques on any given Tuesday, and how little of it is discoverable:

• Which brands are on the shelf today, versus which are listed on a website that was last updated in 2023. • Which specific models, in which sizes, in which colorways. • The frame's real dimensions — lens width, bridge, temple length, and the number almost nobody publishes, total front width. • Which face shapes and head widths each frame actually flatters, which the optician can tell you in four seconds and has never once written down. • What's arriving next month, and roughly when. • Which lens options — polarization, gradient, photochromic, mirrored, custom tints — are available on which frame.

Every one of those data points exists. All of them live in someone's head, a spreadsheet, or a rep's email. None of them are published in a form a machine can read.

III. The AI can only answer with what somebody published

Here is where the problem gets sharper, and where it becomes a strategic opportunity rather than a complaint.

Buying behavior has already moved. A serious shopper no longer starts by walking a neighborhood — he starts by asking. He opens an AI assistant and types something like: "Which shops near the Upper East Side carry Kuboraum in a wide fit, 160mm or more, and what do they have in stock?"

It's a perfectly well-formed question. It's also, right now, unanswerable.

Not because the technology can't handle it. Because the information was never published. An AI engine is a retrieval and reasoning layer over a corpus. It can compare, rank, filter, and recommend with real sophistication — but it cannot conjure an inventory that no one has ever made machine-readable. Ask it, and it will give you a list of well-reviewed opticians and a suggestion to call ahead. Which is precisely the pre-internet answer, delivered by a very expensive computer.

The consequence is that the most qualified, highest-intent, most price-insensitive customer in the entire funnel — someone who knows the brand, knows his measurement, and is ready to buy today — cannot be routed to the store that has his frame sitting on a shelf twelve blocks away. That is not a small leak. That is the whole business.

IV. The size of the thing

To calibrate: the U.S. optical industry was estimated at $65.6 billion in 2023, spread across roughly 44,850 brick-and-mortar optical retail locations. Some 93% of American adults wear some form of eyewear, and non-prescription sunglasses turn over roughly every nine months — a replacement cycle fast enough to make discovery a recurring event rather than a once-a-decade one. Now overlay the structure of the market. Outside the chains, this is an industry of independents: single-location and two-location operators, owner-run, inventory-constrained, allocation-dependent, and competing not on price but on curation and fit expertise.

That combination — high consideration, physical fit requirement, fragmented independent supply, zero machine-readable inventory — is the exact profile of a market where a discovery layer creates enormous value out of nothing. Nobody has to manufacture anything new. The frames already exist. The customers already want them. The only missing component is a way for one to find the other.

V. What it looks like if the shops publish

Now run the counterfactual. Suppose these boutiques posted — natively, continuously, in a format built to be read by AI engines rather than buried in a feed — the things they already know:

Dimensional truth. Not "medium," not "unisex," but the actual numbers: front width, lens width, bridge, temple. Frames are engineered objects. Faces are measurable. The industry's refusal to publish the one specification that determines fit is genuinely remarkable.

Fit intelligence. Which head widths and face shapes a model actually suits — the knowledge the optician deploys instinctively on every customer and stores nowhere.

Live and inbound inventory. What is on the shelf now, and what is arriving, and when. The September shipment announced in August, to the specific people who wanted it in July.

Style and identity signal. This is undersold, and it matters more than the spec sheet in a category this expressive. Who wears this house, and what does wearing it say? Which cultural register does the brand occupy — architectural, aristocratic, athletic, avant-garde? A customer choosing between Kuboraum and Barton Perreira is not comparing hinges. He is choosing between two statements about himself.

Lens and customization options per frame, so the customer arrives knowing what's possible instead of discovering the limits at the counter. Publish that, and something structural changes. The shop stops being a place people wander into and starts being an answer people are sent to.

The visit still happens. That's the point — nobody is arguing that eyewear should be bought sight-unseen, and the try-on is genuinely irreplaceable. But the visit changes character entirely. The customer arrives pre-matched instead of speculative. The optician's hour — the true scarce resource in this business, far scarcer than inventory — gets spent on someone with an 80% chance of buying rather than someone working through a wall of frames by trial and error. That is the difference between a boutique that survives and one that compounds.

VI. Why the platform layer matters more than the app

A dozen startups have built virtual try-on and PD-measurement tools. They haven't solved this, and the reason is instructive: they optimized the fitting step while leaving the finding step untouched, and they built for the brand rather than for the fragmented independent retailer who actually holds the inventory. What the market needs is not another try-on widget. It's a publishing and demand layer where the independents can:

Be inside the AI's answer, not competing for rank in a social feed. Feed placement is rented, volatile, and adversarial. Being the cited source of an answer is durable, and right now the position is uncontested because almost nobody in this category has claimed it.

Own the customer relationship and remarket to it. Today, the man who wanted a 160mm Kuboraum in July is unreachable in September. In a world where the shop owns its list, he is a notification, not a coincidence. The single highest-ROI marketing asset in specialty retail is the person who already came in and left empty-handed — and it is currently thrown away at a rate approaching 100%. Refer across the network and get paid for it. Independent opticians already send customers to each other constantly, out of professionalism and goodwill, and are compensated exactly never. A multi-level affiliate structure turns that informal courtesy into a functioning referral economy — the shop that doesn't have your frame can profitably route you to the one that does, which is better for the customer, better for both shops, and impossible under the current arrangement. Keep the data. Every search, every fit match, every near-miss is signal about what to order next season. On a marketplace or a social platform, that signal enriches the platform's model. It should be enriching the buyer's.

VII. The pattern generalizes — and that's the real thesis

Eyewear is a clean case study, not a niche one. The same structure appears wherever four conditions coincide:

1. The product must physically fit a body, so a store visit is genuinely necessary.
2. Selection is curated and shallow — a boutique carries dozens of relevant SKUs, not thousands.
3. Supply is allocated and intermittent — shipments arrive, sell through, and vanish.
4. The retail base is fragmented into independents with expertise but no data infrastructure.

Run that filter and the list is long: bespoke and made-to-measure tailoring. Footwear with true width sizing. Fine jewelry and bridal, where sizing and stone selection demand the counter. Wigs and hairpieces. Hearing instruments. Performance running shoes fitted to gait. Specialist cycling and ski equipment. Every one of them sends customers wandering into stores hoping, and sends stores waiting for whoever happens through the door.

In each case, the inefficiency is identical and the fix is identical: the seller knows things about the product that would qualify the buyer, and has no mechanism to publish them where the buyer is now asking.

VIII. The bottom line

The instinct in retail for fifteen years has been that technology's job is to replace the store visit. In fitted, curated, expert-led categories, that instinct has been consistently wrong, and the results show it — high return rates, commoditized margins, and the slow erosion of exactly the specialist knowledge that makes these categories worth participating in.

The correct job for technology here is the opposite: make the store visit deterministic.

Not "come in and see what we have." Instead: your face is this wide, your style reads this way, and we have four frames in the case right now that meet both criteria — plus two more landing on the 14th that we've set aside for someone exactly like you.

That message is worth more than any advertisement, and every one of these shops already possesses everything required to send it. They have the inventory, the measurements, the expertise, and the taste.

What they don't have is a place to put it where the AI is looking.

That's the gap. It's not small, it's not confined to eyewear, and it will belong to whoever fills it first.

IX. And here's the fun part

Which brings us, rather neatly, to YESODI — because building this layer requires exactly two things at the same time, and almost nobody has both.

You need the publishing and discovery side: a place where a boutique posts what it actually has, described the way an AI can read it, so the shop shows up inside the answer instead of somewhere on page four of a feed. And you need the fit side: something that knows the human being on the other end — face, frame, proportion, style — well enough to turn "we carry Kuboraum" into "we have three frames in the case right now that were basically made for your head."

YESODI is built for the first. And the second is already living inside the app, because DressMe Online is being built right in — an AI stylist that takes a single photo, no measuring, no forms, no account, and renders looks directly onto the actual person. Point that same capability at a wall of frames and the whole problem inverts: instead of a customer trying on a hundred pairs hoping one lands, the shop can say, we ran your photo, we know your proportions, here are the four that work — and one of them is being held at the front counter with your name on it.

Add the rest of the YESODI machinery — remarketing so the September shipment finds the July customer, five-level affiliation so the optician who doesn't have your frame gets paid for sending you to the one who does, and data that stays with the business that earned it — and the layer isn't theoretical anymore. It's just switched on.

The store visit stays. The try-on stays. The optician's eye stays, thank goodness. All that disappears is the wandering.

Not bad for a pair of sunglasses 😎