Bonaca is a revenue-decision copilot for professional yacht charter fleets. It reads bookings, seasonality and competitor pricing, and tells an operator when to hold a price and when to let it go — with the reasoning shown. The operator decides. Bonaca makes the call legible.

Early-stage product I'm developing, currently in concept and design. Role: market research, product strategy, information architecture, interaction design, visual identity. 2026, ongoing.

Croatian charter boats are booked about sixteen weeks out of fifty-two, so an empty week isn't a gap — it's the margin. When a boat isn't booking, the reflex everywhere is the same: drop the price and hope. Some of those weeks would have filled anyway; nobody knows which.

Croatia peaks by mid-June, Greece in August. Croatian operators make their hardest pricing calls earliest, with the least information.

Existing tools are booking systems, not decision systems. The largest compares competitor prices and leaves the call to you. In adjacent markets like short-term rentals, revenue management is a mature category — nobody was doing it for charter supply. That gap is where Bonaca sits.

Every tool in this space is a dashboard: a screen full of numbers that leaves the decision to the user. I designed against that. Bonaca has four states, each matched to a moment in the operator's year.

Brief — this week. A decision list, not a dashboard. Each boat is a row with the recommended price, the move, and the few numbers that justify it. Accept, adjust or reject; the system remembers why.

Panic — when the market drops. One boat, two futures: drop now for a certain smaller number, or hold for a larger uncertain one. The gap has a name — the price of panic.

Season — the year in reverse. Where money was left on the table, and where the system's own recommendations were wrong.

Owner — read-only. The same data in plain consequence, for the person whose boat sat empty.

Underneath all four: the operator can tell the system what the model can't know — "the festival moved", "my German guest always takes week 28" — and Bonaca shows how much that changed its own recommendation. An AI that listens back.

I reached for "78% likely to fill". Under stress nobody feels a percentage — they ignore it or over-trust it. So confidence became a sea state: a calm line, a ripple, chop. Shape carries the meaning, not colour, so it survives a bad screen and a colour-blind user. Low confidence became a feature: "not enough signal yet, I'd rather tell you Thursday" earns more trust than false precision.

I reached for urgency — red, alerts, visual weight. Adding urgency to a scared person doesn't produce action; it produces a worse decision. So the crisis screen is the calmest screen in the product: no red, more whitespace than anywhere, one decision, the trade-off named in money — and "drop anyway" sits right there, unjudged. A tool that overrides you once doesn't get opened twice.

My instinct was to keep the system's misses quiet. Instead, Season lists the calls it got wrong and what they cost — plainly, in the first person, without excuses. A tool that's never accountable never gets trusted twice.

Token-based design system, shared components across every screen, two users with adapted language — trade terms for the operator, plain consequence for the owner. The wordmark is the product's own settled sea-state glyph: identity and interface share one form. An accent in exactly three places, money always in ink, nothing ever red.

The system is small on purpose: four colour tokens with WCAG AA contrast documented, one type scale with tabular figures for every number, and a handful of components with their states and a rule for when to use each. Colour signals action, never performance.

This is the public version; happy to walk through the rest in a conversation.