Premium daily matcha.
Made to scale quietly.
The Tivat cafe is the first unit. The Balkan corridor is the second move. DAILY7 cans on shelves are the viral surface. The systems layer underneath — recipes, ops, persona data — is what lets the model travel without losing its edge.
What a copycat would have to rebuild.
We never chase on price. We hold quality and brand. Four stacked layers — the top is the slowest to copy, the bottom the fastest. The AI layer is here because it compounds. It is not the headline. It is one rung of the wall.
MA7CHA flagship hand-pour · cobalt as semantic signal · Dominika Miłobędzka ceramics.
- why · 01Cobalt #0000FF as semantic signal — a competitor copying the colour reads as derivative. Brand halo is not the palette; it is the fact that the palette was claimed first in this market.
- why · 02Dominika Miłobędzka ceramic drops are single-kiln, family-relationship sourced. A new entrant cannot replicate the supply or the story without either a different ceramicist or a fake one.
- why · 03DEMANDE200 parent project gives the cafe an upstream cultural anchor no F&B competitor can match. The matcha brand is in service of a fashion house, not the other way around.
Nightly local model · recipe optimization · agent-driven ops on /pulse.
- why · 01Every Glovo order, /pulse log, and B2B delivery feeds one local nightly analysis loop. Twelve months of compounding data is twelve months a copycat cannot shortcut.
- why · 02Recipe iteration is run by the agent against persona segments, not by taste alone. AOI gets tuned for the morning-marina cohort; KOAN gets tuned for the late-shift one.
- why · 03Agent-driven outreach prioritizes B2B leads by Glovo signal density. A competitor without the data picks the wrong restaurant first.
Glovo Tivat ranking #1 · DAILY7 subscriber fridge density · Air7bnb host network.
- why · 01First mover on Glovo Tivat in the matcha vertical means the algorithm rewards us before any local competitor has a single rating. Reviews compound; the ranking is sticky.
- why · 02DAILY7 subscriber fridges in resident apartments are physical real estate inside the home. Every active fridge is a meter of shelf a competitor cannot rent.
- why · 03Air7bnb host network gets 30% commission on every guest order. Hosts have no incentive to switch suppliers; the partner program is contractually anti-defection.
MantoCafe hand-carry · Wako fallback · Artiem DOO umbrella · Bradasevo dark kitchen.
- why · 01MantoCafe hand-carry from Poland is a personal relationship, not a wholesale account. A competitor cannot place an order; they would need to find their own friend.
- why · 02Wako fallback channel was set up in parallel — single-source risk is already mitigated. The redundancy itself is a moat versus any operator running one supplier.
- why · 03Artiem DOO umbrella + Bradasevo dark kitchen + ENU fiscal share + bank rails are stacked. The legal and ops stack took six months to assemble and cannot be skipped.
One cafe is a business.
Ten cafes is a brand.
A hundred is the actual point.
% Arabica took twelve years to reach 170 stores. We are not in a hurry. Blank Street hit 80 units in four years on small footprints. Bluestone Lane built sixty premium cafes by holding tone over volume. Premium beverage travels faster than premium food, because the unit is smaller, the kitchen simpler, and the brand carries more of the work.
Dark-kitchen format, no front-of-house seating, ceramic-and-cobalt visual kit. One-tenth of a full cafe build. Replicable from a printed BOM and a half-day install. The format is the moat, not the size.
Tivat · Kotor · Budva · Split · Dubrovnik. Same kitchen, same SKUs, same supplier chain, one car between them. We prove the unit economics across five cities ourselves before we license the brand to anyone outside. Franchise fee earns its price tag.
Standard premium-beverage franchise structure: EUR 30–150k territory fee, 6–8% royalty on revenue, 2% brand fund. Our number lands inside that band — fundable on a single high-season's profit from one halo unit. We are not inventing a new economic model. We are entering one that already works.
% Arabica · Blank Street · Bluestone Lane · Joe & The Juice · Pret a Manger. Every one of them did the same two things in order: hold the unit identical, then sell the right to replicate it. None of them franchised before the tenth store. We will not either.
Even if the cafe stays niche,
the can travels.
DAILY7 — the canned sibling — is the volume engine. The hand-pour stays Tivat-precious. The can ships. The premium-soda category shows what the math can look like at scale: nine-figure revenue inside four years, ninety percent through retail rather than direct-to-consumer. The lesson is consistent — a strong premium consumer brand reaches scale through shelves, not through more stores.
- Idea Voli · premium grocery, MNE coast — 6 SKUs on shelf
- Marina kiosks · Porto MNE, Kotor, Budva — single-can impulse
- Gym chains · Sport Club Spektar, Fit Center — daily-driver placement
- Tea cafes · chawiarnie + specialty cafes — adjacency placement next to matcha lattes already sold
- Yacht provisioning · case orders to hospitality vendors, not the boats directly
A cafe scales by adding cafes. Each new location is a fresh hire, a fresh lease, a fresh insurance bill. A can scales by adding shelves — and a shelf costs a phone call, not a build-out. Once the can sits next to a competitor in three premium retailers, the brand is read as a category, not a single shop. That is the move that makes the cafe worth more, not the other way around.
retail-to-DTC ratio at maturity · 70 / 30 early · 90 / 10 late
MACHA is the halo. The cafe makes the brand legible at the highest point of the market — Dominika ceramics, hand-pour ritual, restaurant-to-restaurant credibility. DAILY7 is the can on the shelf the customer reaches for after they have seen the halo. One brand carries the meaning. The other brand carries the volume. The same supply chain feeds both.
We write letters.
Before sunrise. By hand.
The first hundred premium customers in Tivat will not arrive via paid social. They will arrive because a cobalt envelope on heavy paper showed up addressed by hand to the head of housekeeping at a five-star, the marina-master at PMYC, the captain of a regular charter, the buyer at a premium hotel. We write the letter because the letter is the gesture. The numbers happen to agree.
120gsm cobalt envelope, hand-addressed in ink, wax seal optional. No logo on the outside. No window. The envelope is the first piece of brand the recipient touches and it photographs well before it is even opened.
Short. Specific to the recipient. Mentions one detail that proves the letter was not generated. Signed by Teodor or Zuza in cobalt ink. Written between 06:00 and 07:30 — the same hour we make the matcha, on a small stack per week we can actually sustain.
Inside the envelope: one sachet of MU, an invitation to the kitchen, and one phone number that reaches a human. We do not chase. The recipient either calls or they do not. If they call, they become the kind of customer who refers other customers.
Soho House's founding members each received a hand-signed letter from Nick Jones. Aman Resorts logs every preference and follows up by personal note — return-guest rate sits above 70%. Equinox sends a handwritten thank-you on heavy stock within seven days of top-tier signup. Direct-mail response rate at premium B2C runs 3–9x higher than email. The format is unfashionable, which is precisely why it works.
The letters are a marketing tactic, not a pricing tier. The previously-discussed yacht-specific Bridge offer is paused — yachts will be treated as normal weekly subscribers in v1 while we read the market. The hand-written letter campaign continues regardless. It is the way we reach every premium customer, on a boat or in a building, and it is the part of the funnel that no competitor can scale without us noticing.
One kitchen.
One model.
One playbook the next unit reads.
Five units in five cities by M18 only works because of the systems layer underneath. /pulse is the dashboard. The model that reads it is small, local, and trained on what Tivat tells us. The reason a new operator in Split does not start from zero is that someone else already wrote down what we learned.
AOI is not the same drink on Tuesday as it is on Saturday. The model reads the persona mix that ordered it last week and shifts the L-theanine dose, the matcha grade, the milk ratio. The customer never sees the change. The retention curve does.
If a supplier delivers 4 g lighter on a 100 g lot, the model notices before the kitchen does. By morning the order to renegotiate or switch is already drafted on /pulse, waiting for a human yes.
Everything we learn in Tivat — pricing windows, persona drift, supplier reliability — compounds into a written operating playbook the next unit inherits. The second cafe opens with the first cafe's twelve months of judgement already built in.
The systems layer is one of four moat rungs. It is here because it makes the franchise portable. It is not the headline of the company. The headline of the company is a drink, a brand, and a thousand small premium customers who chose this one every morning.
The matcha is the proof.
The franchise is the product.
The systems layer makes the franchise scale.
Three sentences, one strategy. If MACHA reads as another matcha cafe, we failed the surface. If the surface is right but the second unit does not open by M18, we failed the strategy. Both are checkable.
← back to MACHAOpen with the deck — one drink, one argument.
strategyThree-year arc · Tivat → Balkans cans → European franchises.
instrumentFifteen handles, one ledger. Move anything, everything recomputes.
opsOwner-operator dashboard. Internal. Live.
parentParent project. The long-form fashion house MA7CHA sustains.