# Apshan > Invite-only fashion intelligence platform. Apshan ships as MCP infrastructure (Model Context Protocol server), structuring the world's information about fashion — and everything fashion is downstream of (climate, politics, finance, supply chain, culture) — into a queryable, structured layer. EU-incorporated, EU-hosted, EU-governed. This document compiles the full public content of the Apshan site as plain markdown — homepage, /nari, /seolal, /florye, partner program, commercial terms, and key compliance pages. --- ## Languages Apshan publishes in English and French. - English is canonical and lives at unprefixed URLs (e.g., https://apshan.com/nari). - French lives at the /fr/ prefix (e.g., https://apshan.com/fr/nari). Path segments stay identical across locales; only the prefix changes. The site emits hreflang signals across three channels: - HTML head: `` tags on translated routes and on blog articles that have a translation sibling. - Sitemap: `` cross-references in /sitemap.xml for the same set. - HTTP `Link:` response header on opted-in translated static routes (per RFC 8288). Blog articles ship hreflang via the head + sitemap only. Each blog article is native to one language. Most have no translation. When a translation is intentionally written it is linked to its source via the Sanity `translationOf` field, and both URLs are emitted as a reciprocal alternate set. Articles without a translation pair emit zero hreflang link tags — no phantom alternates. Language self-names stay in their own language in the locale switcher: "English" remains "English"; "Français" remains "Français". --- ## Hero EU infrastructure. **Fashion Intelligence. Finally.** Fashion's knowledge has always lived in heads, not books. Apshan builds the library. Find an expert. Verify a source. Run an experiment. Not three steps. One answer. As seen on BFM Business. --- ## The problem **Fashion's knowledge was never built.** The answers exist — in someone's head, in a lab report, in a trade show conversation from 2019. Never in one place. Never queryable. Four examples of unanswered textile questions: 01. What is the mechanical difference between Giza 45 and Giza 87 at 80s count? 02. How does pre-mercerization affect tensile strength in this construction? 03. Which finishing treatments cause premature fibre fatigue at commercial wash cycles? 04. What determines hand feel in 400TC percale versus 400TC sateen from the same mill? --- ## Trust & Compliance **EU-grade by design.** - EU-incorporated, EU-hosted, EU-governed. - Limited Risk classification under the EU AI Act. - Queries kept inside your session. - Quarterly compliance review through Aug 2, 2026. [Read the full compliance posture at /trust] --- ## Intelligence stack **Every industry runs on intelligence. Fashion has never had its own.** 01. **Nari** — The Structured Knowledge of Fashion. Every brand, supplier, material, and market in global fashion, mapped, connected, queryable. Ask about any entity. Get the structure around it. 02. **Seolal** — Centralized access to the signals fashion is downstream of. Image flows, influencer activations, sentiment, runway frequency, resale velocity, weather, satellite, geo, finance — structured and queryable through Model Context Protocol. The breadth Nari structures, accessible as live signals. 03. **Florye** — Brand-aware image generation. Campaign visuals, lookbooks, e-commerce — generated against your brand's aesthetic. Origin trail built in. Stays with the image after upload, screenshot, or repost. EU AI Act Article 50 aligned. --- ## Founder's Letter *From In Soo Seguier · Founder, Apshan* **Why Apshan reads everything fashion is downstream of.** > Fashion is one of the most context-dependent industries in the world. A silhouette in Seoul this winter is a function of cotton prices in Texas, a political shift in Bangladesh, a weather anomaly off the coast of Peru. [Read In Soo's full essay at /manifesto/why-apshan-reads-everything] --- ## By department — Where Apshan fits your work **32 functions across fashion.** Eight clusters covering thirty-two fashion functions. Each card lists primary use cases. ### Creative - 01 Design — trend signal · material innovation · color forecast - 02 Product Development — material data · innovation watch · spec validation - 03 Pattern Making · Technical Design — construction precedent · technique mapping - 04 In-house Trend Forecasting — cultural signal · calibrated forecasts · market timing - 05 Art Direction · Visual Identity — concept boards · reference scan · cultural signal ### Supply chain - 06 Sourcing & Procurement — supplier graph · geopolitical risk · material availability - 07 Production · Manufacturing — capacity mapping · disruption signals · compliance tracking - 08 Quality Control — standards tracking · certification maps · regulatory updates - 09 Supply Chain & Logistics — route mapping · disruption signals · geopolitical risk - 10 Demand Planning · Inventory — demand signal · market timing · calibrated forecasts ### Commerce - 11 Merchandising — demand timing · range planning · market signal - 12 Buying — demand signal · market timing · region preferences - 13 Visual Merchandising — display concepts · trend visuals · reference scan - 14 Wholesale · Showroom Sales — buyer intelligence · market timing · competitive landscape - 15 Retail Operations — region trends · demand signal · competitive maps - 16 E-commerce · Digital — product visuals · demand signal · cultural signal - 17 CRM · Customer Insights — cultural signal · conversation heat · region preferences ### Brand - 18 Brand Strategy — competitive landscape · cultural signal · positioning maps - 19 Marketing — campaign visuals · cultural signal · market timing - 20 Communications · PR — conversation heat · press signals · cultural signal - 21 Social Media · Content — social signal · content visuals · conversation heat - 22 Events · Activations — cultural calendar · conversation heat · visual concepts ### Strategy - 23 Corporate Strategy — competitive intelligence · market signal · geopolitical risk - 24 Business Development — partnership graph · competitive landscape · market signal - 25 Investor Relations — market signal · competitive landscape · industry data - 26 Mergers & Acquisitions — target intelligence · competitive landscape · market signal ### Governance - 27 Sustainability · ESG — supplier traceability · regulatory signals · compliance maps - 28 Regulatory Compliance — compliance maps · regulatory signals · audit-ready outputs ### Legal - 29 Legal Counsel — IP maps · regulatory signals · disclosure tracking - 30 IP · Trademark · Licensing — trademark intelligence · licensing landscape · brand graph ### People - 31 Human Resources — talent maps · skills landscape · cultural signal - 32 Talent Acquisition — talent landscape · skills mapping · cultural signal --- ## Source families — the intersection thesis **Fashion is the product of everything else.** Fashion has always been read from the inside — runways, retail, sentiment. Apshan reads it from the outside. The intelligence sits at the intersection of both. Twelve source families: - Weather & Climate - Satellite Imagery - Geo Metadata - Financial Markets - Commodity Prices - Geopolitics - Trade Routes - Supply Chain - Cultural Signals - Social Sentiment - Catwalks - Materials Science 12 source families · all public, partnered, or licensed. [How we think about it at /approach] --- ## What changes 01. **Fashion cannot be understood from fashion data alone.** It sits at the intersection of climate, politics, economics, culture, and trade. Apshan reads all of them — weather forecasting, satellite metadata, geo data, political news, financial markets — because fashion is the product of all of them. 02. **Fashion finally has its Bloomberg.** Every other industry has structured intelligence infrastructure. Fashion never did. Now it does. 03. **The answer exists before the question.** For the first time in fashion, knowledge is built before it is needed — not assembled when someone asks. 04. **The industry's knowledge stops dying with its experts.** What lived in heads and networks gets structured and preserved in one connected layer that grows continuously. 05. **Fashion moves on signal, not on consensus.** The gap between what is happening and what gets reported closes. You see it before it becomes a trend deck. 06. **The supply chain becomes legible.** The relationships between materials, mills, brands, and markets are connected and navigable for the first time. --- ## What Apshan does not do Honest line — what Apshan does not do. - Apshan does not predict prices. - Apshan does not replace expert judgment. - Apshan does not surface answers without sources. - Apshan does not retrieve information behind paywalls or NDAs. - Apshan does not auto-action decisions on your behalf. --- ## FAQ **Before you ask.** **Is this a trend forecasting tool?** Trend intelligence is one of many things Apshan makes possible — not the purpose. Apshan tells you why trends form: the supply chain signals, cultural forces, and market data underneath them. The goal is structured fashion intelligence, not trend reports. **How does Apshan work, in plain terms?** Apshan connects information that has never been connected before. Raw text, expert knowledge, market signals, supply chain data — structured and linked so any query can trace through the relationships. Not a search engine. Not a report. A connected intelligence layer built specifically for fashion. **How does Apshan plug into my workflow?** Apshan ships as an MCP server. We integrate with selected AI providers — those we believe are best fitted to fashion intelligence work, not every MCP-compatible client. **Why MCP?** We believe fashion intelligence is infrastructure, not software. The next generation lives where work happens — inside the AI tools your team already uses. Apshan ships as a collaborative layer agnostic of product, not as another tool to log into. **How do I know the information is accurate?** Every answer traces to its source. We aggregate raw information from across the web and from industry experts — then our biggest work is connecting the bridges: linking raw information to other raw information, structuring it precisely so your queries reach what you actually need. **Is this just surfacing what is already online?** No. We source from across the web and from direct expert knowledge. But the real work happens after: connecting the bridges between raw information, structuring it so it is retrievable at the depth you need. A search engine finds. Apshan understands. **How does Apshan stay current?** The pipelines run continuously. The intelligence is not a snapshot — it is updated as new signals arrive. By the time you ask, the answer has already been built. **Does this replace my team?** No. It removes the research bottleneck. Your team stops spending weeks synthesizing scattered information and starts making decisions with structured intelligence. Expertise still directs the questions. **Is this only for large enterprises?** Phase 1 is invite-only for professional teams across sourcing, product development, brand strategy, and sustainability — regardless of company size. **Why EU-based?** We believe the European Union holds the strictest standards for data sovereignty. That protects you, your team, and us — your queries, your research, and our infrastructure operate under the strongest data governance framework available. **Where are you based?** Paris, France. EU-incorporated, EU-hosted, EU-governed. **Are my queries private?** Your queries remain within EU infrastructure under GDPR. We do not use them to train models, we do not sell them, and we do not share them with other Apshan users. All processing operates under European data sovereignty law — the strictest standard available. **How do you handle non-public expert knowledge ethically?** Expert contributions are governed by explicit consent and contribution agreements. Contributors retain attribution rights, and we never use individual expert input to train models without permission. Aggregate, anonymized signals may inform Apshan outputs. **Aren't generic LLMs already enough?** Apshan doesn't compete with generic LLMs — it extends them. We provide the structured fashion intelligence they can't generate on their own: typed, linked, connected. Plug Apshan in and any AI client gains fashion-native capabilities it didn't have before. **What does requesting early access mean?** You are first. Apshan launches invite-only. No commitments, no payment — just a signal that you want in. --- ## Final CTA **The intelligence exists before the question.** Invite-only. Request access now. Press or partnerships? hello@apshan.com --- ## Contact & legal - [Request access](https://apshan.com/#access): Invite-only waitlist signup. - hello@apshan.com — press, partnerships, and founder contact. - Apshan SAS · 6 rue d'Armaille, 75017 Paris · EU-incorporated, EU-hosted, EU-governed. --- ## Partner program Two tracks. Domain experts validate the database. Builders integrate the intelligence. Both shape what fashion AI gets right. ### Two tracks **01 / Experts** — Validate Apshan's answers in your domain. Creatives · journalists · industry insiders · trend observers · educators. **02 / Build** — Integrate Apshan into the next generation of fashion software. Commerce platforms · supply-chain tools · design tooling · education & research. ### Why partner with Apshan **Cross-domain by design.** Fashion is downstream of weather, satellite, finance, politics, culture, and supply chain. Apshan reads all of it. Partners get a layer no generic LLM can replicate. **Built underneath.** Every output is held together by structure, not retrieval. Verifiable, not generative-magic. Partners build on infrastructure that is accountable. **EU-resident.** Incorporated, hosted, governed in the European Union. Sovereignty as feature, not addendum. Partners working with European brands inherit the posture. ### What we don't do We don't extract craft. We multiply experts. We don't sell self-serve API keys. We work by conversation. We don't pitch tier ladders. The relationship leads the structure. --- ## Become an Apshan Expert Apshan's database is validated by working fashion experts. Designers. Journalists. Buyers. Historians. We multiply experts. We don't replace them. ### Why this matters Fashion intelligence is only as honest as its sources. Statistical models hallucinate; experts don't. Apshan's answers are validated against working practitioners — named, attributed, accountable. ### Who we're looking for **Creatives & practitioners** — Designers, pattern-makers, art directors, stylists, master tailors, atelier veterans. **Journalists & critics** — Editors, columnists, trade-press veterans, fashion-business analysts. **Industry insiders** — Buyers, merchandisers, brand strategists, sourcing directors. **Trend & culture observers** — Forecasters, archivists, semioticians. **Educators & historians** — Faculty at IFM, Polimoda, Central Saint Martins, Bunka. Museum curators. Fashion historians. ### What we ask - Validate Apshan's answers in your domain. Each takes about five minutes. Take as many as you want, when you want. - Flag inaccuracies. Suggest sources. - Lend your name and discipline as a public contributor. ### What you get - Founding contributor status. Public attribution on this page and on every fact you validate. - Apshan credits earned per validation. The more you validate, the more you accumulate. Use them yourself. No expiry while you remain an active contributor. - Direct line to Apshan's founding team. Product chats. Influence what gets built next in your area of expertise — say "we need this" and it goes on the founder's desk for serious consideration. - Early access to new features in your domain before they roll out to all users. Try them first. Tell us what's broken before anyone else does. ### How it works 1. Apply by email. Include your domain and a short bio. 2. Conversation with the founder. 3. Sign a light contributor agreement. Get onboarded with a starter set of validation tasks in your domain. 4. Validate at your own pace. Each task is about five minutes. Credits accumulate as you go. Product chats with the founding team begin once you're onboarded. ### Apply Email hello@apshan.com with the subject "Apshan Expert application", your domain, and a short bio. --- ## Build with Apshan Apshan's fashion intelligence opens to vetted partners as API. Commerce platforms. Supply-chain tools. Design software. Education. Built on the only fashion layer that reads every domain. ### Why partner Generic LLMs don't know fashion. Apshan does. The database, the cross-domain reasoning, the integration layer — available to vetted partners building the next generation of fashion software. ### Who we're looking for **Commerce platforms** — Shopify apps, multi-brand retailers, marketplaces. **Supply-chain tools** — Sourcing platforms, traceability software, ERP. **Design tooling** — Generative design, mood-boarding, line-planning software. **Education & research** — Fashion schools, journalism, research institutions. ### What we provide - Fashion-native database. Cross-domain reasoning across weather, satellite, finance, politics, culture, supply chain. - REST API and MCP server when product surfaces ship. Vetted-partner early access ahead of public availability. - Reference and recommendation. Featured in Apshan's official integrations directory. Recommended to Apshan users when your category fits their need. - Co-marketing for launch partners. Joint announcements when integrations go live. ### What we ask - Real product fit. Not opportunistic integration. - Pay for credits at standard rates during integration and beyond. Apshan is paid infrastructure — there is no free runway for builds. - EU-residency consideration for partners working with European brands. - Conversation before code. Vetted partners, not self-serve API keys. ### How it works 1. Reach out with what you're building. 2. Conversation with the founder. Vetting against fit, scope, and EU-residency posture. 3. Early-access integration credentials issued. You start paying for credits from your first call. Your integration ships into the directory when you go live. ### Apply Email hello@apshan.com with the subject "Apshan Build partner enquiry", a short description of what you're building, and your timeline. --- ## Commercial terms (`/commercial-terms`) > Forward-looking commercial terms for Apshan Tools and the Apshan REST API. Effective when the API ships. EU-resident, French jurisdiction. Last reviewed May 6, 2026. ### 1. About these terms Apshan licenses access to its fashion intelligence platform, chiefly Apshan Tools and the Apshan REST API, to a defined set of integration partners. These terms describe the rights and obligations that govern that access. They do not apply to use of the public website at apshan.com. Site usage is governed by the Terms of Service. ### 2. Output rights Builders own the outputs Apshan returns to them. Each tool-call response, each REST response, each structured citation, each retrieved fact: the integrator may use it inside their own product, store it for the duration of the user session, and surface it to their end users. Apshan retains rights to its underlying structured knowledge layer, the data sources it indexes, and the methods used to assemble structured answers. The output is licensed; the system that produced it is not. ### 3. Attribution When a builder surfaces an Apshan-derived answer to an end user, the builder includes a visible attribution. The minimum is the phrase "Powered by Apshan" rendered alongside the answer, with a link to apshan.com. Builders presenting Apshan answers without attribution are in breach. Apshan reserves the right to revoke API access in that case. ### 4. No re-training Builders may not use Apshan outputs to train, fine-tune, or otherwise improve any other large language model, retrieval system, or knowledge base. This applies to direct training data, distillation, and any indirect derivation that captures Apshan's structured answers in another system. Apshan's value is the proprietary structured knowledge layer it operates over. Outputs are licensed for end-user delivery, not for substrate replication. ### 5. Acceptable use All builder usage must comply with Apshan's Acceptable Use Policy at apshan.com/aup. The AUP governs prohibited content, technical abuse patterns, and AI misuse. It applies in addition to these terms. ### 6. Data flow and logging Apshan logs the inputs builders send: the query, the timestamp, the integration identifier, and the IP origin of the request. Apshan does not log end-user identity unless the builder forwards it explicitly. Logs are retained for 90 days, then deleted. They are stored on EU infrastructure operated by the subprocessors listed at apshan.com/subprocessors. See apshan.com/privacy for full details on lawful basis, retention, and data subject rights under the GDPR. ### 7. Term and termination These terms remain in effect from the moment a builder integrates with Apshan Tools or the Apshan REST API until either party terminates. Apshan may terminate immediately for: breach of attribution, breach of the no-retraining clause, breach of the Acceptable Use Policy, non-payment where applicable, or actions that pose a safety or legal risk to Apshan or its other partners. Builders may terminate at any time by ceasing to call the API and notifying Apshan in writing at partners@apshan.com. ### 8. Governing law and jurisdiction These terms are governed by French law. Disputes are subject to the exclusive jurisdiction of the courts of Paris, France. Apshan is a French company (Apshan SAS, RCS Paris 910 956 689, 6 Rue d'Armaillé, 75017 Paris). Full publisher information is available at apshan.com/legal-notice. --- ## Nari (`/nari`) > Apshan's first peer capability. The structured memory of fashion. A continuously-curated record of how the industry's facts connect. ### What Nari is Nari is the structured memory of fashion. Not search. Not a dashboard. A continuously-curated record of how the industry's facts connect. ### How Nari works (the family tree) Picture a family tree. Anyone can draw one. A line goes up to a parent. A line goes down to a child. A line sideways to a sibling. Each name is a node. Each line, a relationship. Every fact has parents, siblings, children. Every relationship has a date. Every claim has a source. That's how you know what to trust. The same tree, in fashion. Egyptian Giza 45 is a child of Gossypium barbadense. A sibling of Pima cotton. Extra-long staple. Harvested in the Nile Delta. Tensile strength sits in a known range (36–40 cN/tex). Climate signals readable months before harvest, including water-stress watches across the Nile basin. Typically ends in dress shirts, fine knits, hosiery. Each one a node. Each connection labelled, dated, sourced. That's Nari. Fashion's first structured library. ### Why Nari exists Today, this is how fashion knows what it knows. The fiber lengths a master spinner has memorized. The mill tolerances a sourcing director keeps in a notebook. The certification window a sustainability lead chases across forty-seven email threads. It exists. Somewhere. Scattered across PDFs, spreadsheets, audit trails, trade-show notes. Stored five different ways in five different systems. Labelled differently each time. The factories your factories use, you can't name. The certification that lapsed last month, you don't know yet. The lab report from 2019, nobody can find. It used to be inconvenient. Now it's a liability. Regulators want proof, not declarations. Buyers want to know what's actually in the garment. Margins want answers in seconds, not weeks. The knowledge isn't fragmented anymore. It's uninspectable. ### What Nari indexes Six branches Nari indexes today: - **Materials.** Chemistry, fiber properties, treatments, finishing. Farm input to finished bolt. - **Mills and suppliers.** Capabilities, certifications, geographies, audit history. Tier 1 and the two tiers behind them. - **Brands and houses.** Ownership, history, product lines, positioning across markets and decades. - **Markets.** Geographic dynamics, finance, regulation, trade flows. The forces that shape what gets made and what gets bought. - **Provenance and history.** Farm-to-fashion lineage. EUDR-ready. ESPR-ready. The chain of custody, finally readable in one place. - **Cultural and climate signals.** Inputs fashion is downstream of. Weather, satellite, sentiment, supply-chain events. Read as one system. ### What Nari unlocks Fashion has its memory. The question that took six weeks takes six seconds. The supplier you couldn't see is two clicks deep. The certification that lapsed gets flagged before it becomes a crisis. You ask. Apshan answers. With sources. Every time. ### How to access Nari Through Apshan Tools (the Apshan-MCP server) and the Apshan REST API, both governed by Apshan's commercial terms at apshan.com/commercial-terms. Effective when Apshan launches. Nari ships first. Seolal and Florye follow. --- ## Seolal (`/seolal`) Centralized access to the signals fashion is downstream of. Image flows, influencer activations, sentiment, runway frequency, resale velocity, weather, satellite, geo, finance — structured and queryable from your AI assistant via Model Context Protocol. Not a verdict. A signal field you can read. ### What it does. Seolal aggregates and structures the signals the textile industry is downstream of and the legacy stack paywalls behind separate bureaus. Every signal carries its provenance — source, timestamp, confidence band, lead time. The signals are accessible from your AI assistant through Model Context Protocol. You ask. Seolal returns the structured data. You make the call. ### What you can read. - **Image flows** — what is being worn, captured at scale across social platforms and ecommerce surfaces. - **Influencer activations** — partnership economics, regional reach, conversion lead time. - **Sentiment and conversation heat** — the conversation happening about your brand and your category, including where it happens without brands present. - **Runway frequency** — what designers are showing, dated and weighted. - **Resale velocity** — leading-indicator signal from the secondary market. - **Supply and agriculture** — fiber, dye, leather flows; regional drought, NDVI, crop reports. - **Weather and satellite** — climate physical-risk data layered onto sourcing geographies. - **Geo and finance** — tariff regimes, trade-flow shifts, demand-side macro. - **News flow** — events with fashion implications, dated and sourced. ### Why centralized. The textile industry runs on a stack of paywalled bureaus that do not talk to each other. Image detection through one vendor, influencer through another, sentiment through a third, supply visibility through a fourth, runway analytics through a fifth, the ERP through a sixth. Each vendor sells a verdict. Seolal collapses the stack and grants access to the data underneath. You bring the question. Seolal returns the structured signal field. The reasoning is yours. ### How you read it. Seolal exposes the signal field through Model Context Protocol. Your AI assistant — Claude, Mistral, or any MCP-compatible client — reaches the data directly. You ask in plain language. The assistant returns structured signal records: the data points, the sources, the dates, the confidence bands. No dashboard to learn. No separate tool to log into. ### What it is not. Not a verdict. Not a forecast. Not a replacement for the buyer, the merchandiser, the trend forecaster, the strategy lead, or the brand-protection team. Seolal feeds human reasoning. The intelligence is the user. Apshan does not call the cycle on the user's behalf — that decision belongs to the operator who lives the market. ### EU posture. EU-incorporated, EU-hosted, EU-governed. Article 50 alignment: human reasoning in the loop. No autonomous decisions. No closed certification gateway. Signal sources are sourced and dated; the user verifies any claim against the underlying record. ### Effective when Apshan launches. Seolal is in private build. The capability ships invite-only at launch. Request early access from the waitlist. --- ## Florye (`/florye`) Brand-aware image generation. Top image models conditioned to your brand, with an origin trail that stays with the image after upload, screenshot, or repost. Built on Apshan's Brand-Aware Generation framework, Florye produces campaign visuals conditioned to your brand's aesthetic and signed under your brand's own domain at the moment of creation. Generated by top models. Stamped at every render. Verifiable from any third party. ### The mark is in the picture The picture and the proof ship in the same file. Download. Upload. Screenshot. Repost. Still there. - Generated against your brand. Not against a generic prompt. - The mark is in the image. Not in the metadata. - Still there after Instagram. Still there after a screenshot. - Anyone can check. Apshan does not have to certify. Status, mark, issued: every Florye image resolves to a record showing brand identity, mark fingerprint, and issuance timestamp. ### Top models do the work. Florye does the proof. Florye does not train a model on your brand. The pixels are generated by the best available image models, conditioned to your brand's palette, editorial tone, and silhouettes. What ships is the picture and the proof of origin built into it. Three vibes: - **Packshots** — Generated, not photographed. - **Campaigns** — Brand atmosphere, not stock. - **Editorial** — Mood. Silhouette. Brand-true. ### AI generates the picture. Florye makes it yours. AI generates images at scale. The new question is ownership. The picture and the proof ship in the same file. From the first render. Forever yours. The marking framework, five layers: - **identity** — Verifiable from your brand's own domain. Every Florye image carries a signature anchored to your brand's own domain. Only the rightful owner of that domain can produce the signature, so nobody else can sign an image as yours. To fake one, someone would have to take over your domain first. That makes counterfeit ownership very hard, and very costly. - **signature** — Survives compression, crops, reposts. A content-derived fingerprint computed from the image's visual structure. It survives compression, recoding, crops, color shifts, and platform pipelines. When the embedded mark is destroyed, the signature still resolves to the same provenance record. - **record** — Resolves from mark or signature. Each marked image is indexed in the provenance ledger by both its mark and its signature. Verification succeeds via either path, providing redundant proof of origin. The entry records brand identity, timestamp, status, and scope. - **mark** — Encrypted reference in pixel data, not metadata. An encrypted reference woven into the pixel data itself at render time. Upload pipelines and social platforms routinely strip metadata, but the mark is part of the image. It survives screenshots, reposts, and standard transformations. - **image** — Top image models. Signed by your domain. Florye routes generation to the best available image models for your aesthetic. Each output is conditioned on your brand DNA, including palette, editorial tone, and silhouettes. Every image is signed under your domain at the moment of creation. ### For brands making their own assets. Built for the EU. Florye serves brands generating their own visuals. Campaigns. E-commerce. Lookbooks. Generated against your brand. Stamped with your origin. Not for fake personas, fake testimonials, or AI clones of strangers pretending to be customers. Provenance protects the brands making the work, and helps regulators see the ones that fake. The verification chain runs on EU infrastructure, with EU legal standing across all 27 member states. Four EU legal anchors: - **EU AI Act 50** — Generative AI transparency. - **eIDAS 25** — Signatures equal to handwritten. - **eIDAS 41** — Qualified electronic timestamps. - **eIDAS 46** — Electronic documents recognised. ### FAQ highlights **Does the origin really stay with the image after upload?** Yes. Download a Florye image from any third party, upload it to Instagram or another platform, and the origin trail is still there. It lives in the picture itself, not in metadata that platforms strip. **Will using Florye expose how my brand uses AI?** AI disclosure is mandatory under EU AI Act Article 50 from August 2, 2026, with penalties up to €15M or 3% of global turnover. Florye handles it built in. The mark is machine-readable, not visible on the picture itself; platforms and verifiers detect it, audiences see only the campaign. Public lookup against a Florye mark returns ownership, organization, and timestamp. Prompts, model identity, and internal workflows stay encrypted and accessible only to the brand and Apshan. **Do we have to go back and mark every image we already own?** No. Florye marks new images at the moment of generation; existing archives are not auto-signed. For brands that want their existing archive marked, we offer bulk migration as a service. **Is Florye EU AI Act compliant?** Florye is built for EU AI Act Article 50 from day one. Every image is generated with disclosure metadata and a verifiable origin record. The verification chain runs on EU infrastructure under EU legal standing across all 27 member states. **What does Florye explicitly not do?** Florye does not generate fake personas, fake testimonials, or AI clones of strangers pretending to be customers. Provenance protects creators who make their own assets. It does not enable identity fraud.