Enoshima runs research-grade, multi-agent simulations to estimate an artwork's price-appreciation potential — every evaluation centered on a specific piece, every computation on your own machine.
Runs entirely on-device Your own LLM via Ollama No data leaves your Mac
Enoshima exists to answer one question, well. Not whether an artist is fashionable — whether this specific piece, at this price, is likely to appreciate.
It models how the market actually moves: through the people who move it. Then it tells you how confident it is, and where the verdict is thin.
For every evaluation, Enoshima simulates dozens of individual agents across five audience segments, then cascades their reactions through the market the way real opinion spreads — buzz building, conviction forming, prices following.
Set the critical frame and confer legitimacy.
Price-makers who read demand before it shows.
Research-driven buyers who move with conviction.
Taste-followers who broaden a market's base.
Social signal that accelerates — or stalls — a wave.
The cascade →
The result is four readings of the same piece: a cascade score from 0–100, the appreciation pathway it would travel, a consensus map across segments, and a fragility analysis showing where the verdict could break.
Illustrative output. Enoshima surfaces its confidence and flags sparse data rather than overstating a result.
Every evaluation centers on the piece itself — its image, medium, dimensions, asking price, and provenance — rather than the name in the corner. Two works by the same hand can earn very different verdicts, and Enoshima treats them that way.
A work-to-work sales-comparison engine derives a fair-value band and an underpriced / fair / overpriced verdict from comparable sales — normalized for size, currency, date, and medium — and tells you how confident it is.
Agents see the actual artwork. Vision analysis of the image feeds directly into how the simulated market reacts — composition, palette, and presence become part of the read.
Point it at an artist's name or a URL and Enoshima enriches the profile from across the web: biography, exhibition history, past sales, and social-media signals — gathered, not typed.
See how the verdict holds up under price hikes, market stress, and competitive scenarios — so you know what would have to change for the answer to change, before you commit.
Enoshima is built for people who treat discretion as a feature. There is no cloud to opt out of, because there was never a cloud. Every computation happens on your own machine.
Runs your own model through Ollama. No third-party inference, no prompts in flight.
Artists, works, and comparables live in a local graph — no external store, no shared index.
No accounts phoning home, no telemetry, no analytics — nothing about your evaluations leaves the device.
Any API keys you choose to add are sealed in the macOS Keychain — handled by the system, never by us.
Enoshima is an island off the coast near Kamakura, and the home of a shrine to Benzaiten — goddess of wisdom, the arts, and all flowing things. The only woman among Japan's Seven Gods of Fortune, she has been revered by artists for centuries, and she is bound to water and to islands.
The muse who presides over wisdom, the arts, and all that flows — value among them.
She is hidden in the mark: a serene profile carried inside a Hokusai wave, present but understated — the way insight sits beneath data, read on a second look. The gold is the value she keeps watch over.
Bring discipline to instinct. Pressure-test a purchase, and understand what you're paying for, before you fall for it.
A repeatable, defensible read on price and demand for the works you handle — and a clearer case to make to a buyer.
Evaluate art as an asset class with confidence intervals instead of vibes, and a record of how each verdict was reached.
Now in private developmentEnoshima isn't public yet. Join the waitlist and we'll let you know the moment it's ready — with early access for the first to sign up.
Enoshima gives you a research-grade second opinion — it states its confidence and flags thin data rather than promising a return. It is not investment advice, and the judgment stays yours.