Unreleased source candidate - Android v0.8.0

MOSTRAmi turns any smartphone into an on-device AI museum guide.

Visitors scan once, lower the phone and talk naturally about the artwork in their preferred language. Answers show source context; Clio accepts interruptions, on-device operation supports privacy and weak connectivity, and explicit cloud processing enables richer real-time interaction.

Download current signed v0.7.0 private-beta APK Try the Web Simulator Open the Pitch Deck See Build Status
For museums See the seven decisions that activate a pilot
15 Private testers
v0.8.0 Unreleased candidate
0.8.0 (21) Source build passing
0 EUR Pre-revenue

The problem YC should care about

Museums are full of silent context. Visitors want instant explanations in their own language, but most venues still rely on static labels, rented audio devices, or generic mobile tours. Cloud AI creates a second problem: camera data, bystanders, latency, and compliance risk.

Visitors want depth now

They do not want to search, scan long wall text, or rent a device. They want the right story at the exact object in front of them.

Museums need modern guides

Institutions need AI-grade interpretation without sending visual data to external services or operating expensive hardware fleets.

Language is the opening

Mandarin, English, Italian, and other languages become a product surface, not a separate audio-guide procurement cycle.

What MOSTRAmi does

MOSTRAmi is a mobile AI guide for museums, galleries, and cultural sites. The current Android build is focused on proving that local inference can create a useful visitor experience on real devices.

Recognize the context

The app uses the phone camera and local model pipeline to understand the cultural object or scene the visitor is exploring.

Explain like a guide

It generates short, conversational interpretation that can adapt by language, age, curiosity level, and available time.

Keep data local

The product direction is privacy-first: reduce cloud dependency, protect museums from visual-data risk, and keep latency low.

Why this is the right first market

Cultural tourism is a strong early wedge because the pain is visible, the use case is repeatable, and the first buyer can be either the visitor or the institution.

B2C path

  • Tourists pay for a better museum visit without waiting for the museum to adopt new systems.
  • Initial pricing can start with paid day passes, city passes, and premium multilingual narration.
  • Chinese visitors to Europe are a high-value segment for the HarmonyOS roadmap because language access, privacy, and premium travel behavior align.

B2B path

  • Museums can buy private pilots for exhibitions, guided routes, accessibility programs, and multilingual visitor services.
  • The product avoids the operational burden of rented audio-guide hardware and cloud-heavy AI deployments.
  • B2B pilots create curated content, credibility, and venue distribution for the consumer app.
Museum activation blueprint

Seven decisions turn one exhibition into a living guide.

MOSTRAmi does not ask an institution to digitize its entire collection or operate an AI stack. We start with one exhibition, 10–20 works, and one named editorial owner.

01

Scope the pilot

Museum provides

One exhibition, 10–20 works, and one accountable editor.

MOSTRAmi unlocks

A bounded visitor experience that can be tested quickly.

02

Open the source material

Museum provides

Catalogues, labels, rights-cleared images/audio, maps, and existing translations.

MOSTRAmi unlocks

A versioned Museum Pack with traceable sources and rights records.

03

Set the editorial direction

Museum provides

Audiences, languages, tone, sensitive claims, pronunciation, and accessibility needs.

MOSTRAmi unlocks

Clio adapts voice, depth, language, and interpretive boundaries.

04

Review and approve

Museum provides

Approval of works, routes, translations, citations, disputed attributions, and hard limits.

MOSTRAmi unlocks

Museum-grounded answers with visible provenance.

05

Test inside the gallery

Museum provides

A walkthrough under real lighting, connectivity, visitor flow, and photography rules.

MOSTRAmi unlocks

Artwork/label scan, Clio Live Q&A, zoom/tap/circle detail exploration, and local/offline fallback testing.

06

Publish the experience

Museum provides

Editorial sign-off, QR/direct-link placement, and launch timing.

MOSTRAmi unlocks

Routes, partner-supplied maps, Scholar Dossier, visit Diary/recap, and pack distribution.

07

Learn and update

Museum provides

Periodic editorial review and content updates.

MOSTRAmi unlocks

Versioning, aggregate anonymous analytics, and new exhibition, critic, accessibility, language/culture, or education packs.

The museum decides what is authoritative. MOSTRAmi makes it conversational.

The institutional return

One approved content layer creates six kinds of value.

The return is cultural reach, a deeper visitor relationship, editorial control, operational leverage, privacy-safe learning, and new distribution and revenue options.

Reach more audiences

Feature evidenceMultilingual narration, natural voice, readable transcripts, accessibility routes, simplified language, and offline/on-device support.

ResultThe same exhibition becomes useful to more visitors without commissioning a separate guide for every segment.

Create deeper visits

Feature evidenceScan, Clio Live, free-form questions, image details, routes/maps, Diary, and Scholar Dossier.

ResultThe visitor moves from passive playback to active, memorable exploration.

Keep editorial control

Feature evidenceMuseum Packs, approved sources, citations, claim limits, review status, signatures/versioning, and visible provenance.

ResultAI interpretation stays accountable to the institution instead of becoming generic model output.

The compounding loop

  1. CONTENT
  2. VISIT
  3. INSIGHT
  4. UPDATE

Reduce repeated production work

Feature evidenceReuse catalogues and labels; structure, translate, adapt, update, and version once across multiple visitor formats.

ResultOne approved source layer supports many experiences instead of another one-off audio guide.

Learn without profiling visitors

Feature evidenceAggregate scan success, dwell-time bands, most-explored works, route completion, and language demand.

ResultBetter content and visitor-service decisions without collecting raw photos, faces, private chats, precise indoor location, or individual profiles.

Open new channels and revenue

Feature evidencePremium museum/exhibition packs, temporary passes, city bundles, school/education packs, critic/accessibility packs, marketplace, and future partner SDK.

ResultApproved cultural content can keep creating value before, during, and after the physical visit.

The museum does not give up its voice. It multiplies it.

What is real today

MOSTRAmi is not a slide-only idea. It is a private Android beta being tested across devices because local AI performance depends heavily on phone hardware, thermals, memory, and battery behavior.

15-person tester network

A private group receives major APK builds and reports behavior across different Android devices, especially where local execution stresses the hardware.

0.8.0 AI Integrity candidate

The August 14 source candidate is v0.8.0 (21). Android tests and the R8 YC build pass, alongside Functions 8/8, token Worker 7/7 and mostrapack 9/9. The new artifact is unsigned and not distributed; signing, cloud canary and exact-device smoke testing remain explicit gates.

Exhibition PoC in scoping

On July 26, founder Fabrizio Degni and Japanese artist Shoko Okumura agreed to scope a proof of concept for a future exhibition. Scope, venue, dates, budget and commercial terms are not finalized, so it is not counted as a pilot, customer, partnership or revenue.

Execution risks and how we reduce them

Device performance

Risk: local AI can be slow on weak phones. Response: test across real hardware, make model choices device-aware, and keep a cloud fallback optional only where needed.

Museum sales cycles

Risk: institutions move slowly. Response: use B2C tourist adoption and small exhibition pilots before negotiating larger venue contracts.

Overbuilding before demand

Risk: technical breadth can hide weak user evidence. Response: freeze expansion, run real museum sessions, measure second-scan behavior and test willingness to pay before iOS, WeChat, marketplace liquidity or new model integrations.

Gate-based release roadmap

By August 31, 2026

Close readiness gates

Sign and test 0.8.0, canary the Gemini 3.6 grounded backend, verify local speech/model integrity and record an unedited exact-build demo.

September-October 2026

Closed beta, Wave 1

Run at least 30 real-context sessions; measure first and second scan, voice reliability, repeat-use intent and one concrete willingness-to-pay offer.

November 2026-January 2027

Extend and harden

Expand the device matrix, publish the measured beta report, secure one written design-partner scope and complete a production source broker plus one museum pack.

February-September 2027

PoC, then release decision

Run one defined exhibition PoC, test a commercial offer, then decide whether evidence supports an open beta or limited public Android release.

Why YC

YC would compress the hardest part: turning a promising build into a fast-learning company.

MOSTRAmi needs the YC push on focus, speed, pricing, pilots, founder network, and investor readiness. The goal is to move from private Beta 1 to paid evidence quickly enough that the company is judged by usage and revenue, not by polish alone.

Founder commitment

  • Solo founder: Fabrizio Degni.
  • Founder-led product and technical execution.
  • Company base: Milan now, San Francisco during YC if accepted.
  • Immediate ask: feedback, pilot intros, and pressure toward measurable traction.