Data out of control
CRM, documents, voice, automations: everything lands in remote SaaS. For an SME, compliance and customer trust become fragile.
CLARYEL · Pitch
CLARYEL
Local privacy-first AI infrastructure for SMEs and organisations — appliance + software + support, on one platform.
Pre-seed / seed. Honest stage — live demonstration is part of this pitch.
02 · Problem
Small and medium enterprises need AI, but cannot (and should not) push sensitive data into third-party clouds without control. Enterprise-grade technical support exists — usually at prices they cannot afford.
CRM, documents, voice, automations: everything lands in remote SaaS. For an SME, compliance and customer trust become fragile.
Premium service and serious technical support exist — but typical pricing excludes exactly who needs it most: small businesses.
AI chat, helpdesk, monitoring, recovery: separate pieces. No local platform that holds them together under the customer’s control.
03 · Who for
SMEs, professional practices, organisations and homes that want useful AI without giving up privacy. The idea comes from the founder’s hands-on AI experience and from running his own business.
04 · Physical CLARYEL BOX
What we demonstrate live is a local CLARYEL Box: models, memory and automations on the customer’s hardware. UI v1 was tested, then unavailable after an architecture change; UI v2.0 by 10 October 2026. 10–13 physical prototypes already ready; installs go through technical support with AI agents (not by hand) — in parallel: support tests, repeatable installs and system tests. No paying customers yet: honest stage.
Three honest tiers: refurbished entry, new build, recurring support.
from €1.000
Built on a refurbished server with new chassis, cooling and PSU. Deep-tested, with warranty.
from €1.500
Brand-new build, with leasing, financing, credit or installments available.
from €30/month
Subscription for technical support of the full stack (monitoring, updates, recovery, RemoteOps).
Live demo
Today’s live scenario: one multi-topic voice query on the local corpus → partially correct answer with a controlled technical error → call an ordinary telephone number → remote AI agent connects, fixes and sets up → the encrypted personal-data volume disconnects automatically (visible).
One spoken question on the local corpus — several topics in a single ask.
A partially correct answer and an intentional technical error — to open support.
Call a normal phone number — not an exclusive portal.
The AI support agent connects remotely, fixes and configures.
On connect, the encrypted personal-data volume disconnects — you can see it.
05 · AI & Data Architecture
One assistant for the user. Behind it: multiple specialised AI models, vector + structured databases, HA, containers. Two data layers: technical/operational vs private customer.
«Support can see the health of the system — not the private life inside it.»
«Поддержка видит здоровье системы — не частную жизнь внутри.»
«Il supporto vede la salute del sistema — non la vita privata al suo interno.»
«One assistant. Multiple AI models. Separate data layers.»
«Один ассистент. Несколько ИИ-моделей. Раздельные слои данных.»
«Un assistente. Più modelli AI. Layer di dati separati.»
«Support can see the health of the system — not the private life inside it.»
«One assistant. Multiple AI models. Separate data layers.»
06 · RemoteOps / support
In the live demo: you call an ordinary telephone number; a remote AI agent connects, fixes and sets up. When support connects, the encrypted personal-data volume disconnects automatically — visible on screen. RemoteOps is the path: voice, remote access, tickets, recovery — without entering the customer’s private layer.
Call a normal phone; the remote AI agent connects, fixes and configures — no exclusive portal.
On remote support connect, the encrypted personal-data partition/volume disconnects automatically — visible. RemoteOps: voice-first (Windows / Linux / macOS), policy, approvals, health, rollback — remoteops.claryel.space.
Accepted into the Deepgram Startup Program: voice-AI credits already available for the support line (tickets + voice). We can offer the same line to other startups free or on credits, without asking for equity.
Talks open with Google: they are waiting for the live commercial site; after approval we expect a larger support/credits package. Status is also tracked in the investor cabinet.
07 · Competition and advantages
Cloud copilots, “AI NAS”, traditional MSPs: pieces of the puzzle. CLARYEL combines local appliance, privacy layers, shared platform and operational support.
| Capability | Cloud AI chat | NAS / home lab | Classic MSP | CLARYEL |
|---|---|---|---|---|
| Data on customer perimeter | No | Partial | Variable | Yes · by design |
| Private layers separate from support | No | Rare | No | Yes |
| Hardware + software + ops | Software only | Limited HW | Ops without local AI | Unified stack |
| SME-accessible price | SaaS seats | DIY | Often high | Explicit goal |
| Platform for other startups | No | No | No | Shared architecture |
08 · Traction & financing
No official paying customers yet. We are our own first customers. Non-dilutive funding already awarded; prototypes and installations are the next commercial proof.
Investor cabinet live: investors.claryel.space — operating indexes, funding publication-approved, no invented valuation. In the hub: Google for Startups talks in progress (waiting on the sales site); Deepgram credits already active for the support line.
Concentration on Box (claryel.com) and identity (id.claryel.com). Other product lines on operational pause.
09 · Business model
Intended revenue from Box appliances, initial consultation, managed support and (over time) platform services. Goal: enterprise-level service at an SME-accessible price. Numeric model detail stays in controlled disclosure.
Public entry offer at €99 (credit toward Box hardware) — sector analysis + consult; path toward purchase or prototype.
Local appliance, commissioning, identity, private layers. Prototype: from €1,000 (min. refurbished) / from €1,500 (new, with leasing/financing). Support from €30/month. Warranty 1–5 years by plan.
System health monitoring, update, recovery, RemoteOps — without access to the private customer layer.
10 · Future / infrastructure draft
First a repeatable fleet in Italy; then country by country. Goal: from proven prototypes to serial supply, with support and recovery.
First scale across Italy with delivery services; then country by country in the EU, respecting local specifics; then the United States.
Strategically we aim at every sector: we start with those who show real interest in the prototypes. The solution is universal — we work across all verticals.
Once prototypes have enough certification and validation for personal use, we open sales to households: price and subscription are designed to stay accessible for families.
After full testing of technical support and our universal hardware profiles / configurations, we can bring user computers, their local networks, NAS and other auxiliary systems (including smart home) onto the same support line and server care: diagnostics, recommendations to replace or upgrade hardware — and, where it makes sense, perform the upgrade ourselves.
11 · Startup ecosystem
Identity, deploy, monitoring, model routing, knowledge and recovery are reused across products. The founder has many vertical ideas — not enough resources to execute them alone. The platform is built for partnership, co-build and sharing the tech-support line.
Local privacy-first runtime.
Voice-first computer ops + tickets.
Managed + community edition.
Shared sign-in across cabinets.
Offer to other startups: a support line as micro-investors — no equity, only collaboration.
We can share our support line (tickets, voice, RemoteOps) wired to the partner’s knowledge base so their customers get help on their product. Often free thanks to Deepgram credits (and, later, Google). We can also hand over a ready-configured stack and know-how at near-zero cost.
Using our line is optional: we help set up a separate tech line for the partner from a few euros per month, with a dedicated number and WhatsApp / Telegram channels.
We act as operational micro-investors: we do not ask for shares. We want collaboration — sharing technology, knowledge and support capacity.
Open to any idea or project on the CLARYEL local infrastructure and to embedding other startups’ services into our product — from a free trial to a full partnership.
12 · Why now / who we seek
We are here for capital, customers and networking with other startups. Stage is pre-seed / seed: honest on traction, clear on direction.
Resources to take prototypes to repeatable installations, strengthen the team and accelerate intelligent RemoteOps — without inventing metrics.
Organisations willing to pilot / install Box, with real feedback on the support cycle.
Teams that build verticals on the CLARYEL architecture, bring go-to-market in Italy/EU, or want a support line (ours or dedicated) in a no-equity collaboration.
13 · Team + contacts
A small, focused team: founder, design & brand, plus a strategic advisor on architecture.
Founder & Product Architect · architecture, product, engineering, operations. Como / Fino Mornasco.
Head of Design & Brand · fashion & product designer, graphic design, apparel & accessories, brand and creative visuals. Former brand director and lead designer (2016–2024, manufacturing): end-to-end company identity and product design. Como / Fino Mornasco.
Strategic advisor · Ph.D. in Computer Science; founder/CTO. Designs and builds complex software products end-to-end (architecture, infrastructure, native apps, UX); GenAI platforms, distributed systems and security at 20M+ monthly active users. Supports product architecture decisions.
14 · Q&A prep
Short, honest answers — no inflated traction or IP.
Not officially. We are the first users. 10–13 physical prototypes already on hand; UI v2.0 installs via technical support (AI agents). Non-dilutive €10k funding already awarded + Deepgram Startup.
UI v1 was tested, but due to an architecture change it is currently unavailable; UI v2.0 will be demonstrated by 10 October. Hardware: 10–13 physical prototypes already ready on hand, waiting for UI v2.0 installation via technical support (AI agents). Architecture (specialised models + two data layers) remains the differentiating core.
No: it is a deliberate strategy. We go broad because the project is unique, infrastructural and complex. Only that way do we reach the final product in the last development stage: we freeze or hide functions not yet ready for operation, replace them with simplified versions, or — as with technical support — with manual operations.
Not yet. Designs not filed. Prototypes = internal developments.
Today by phone — operational but not yet intelligent enough. RemoteOps + privacy boundary (system health ≠ private life) is the path. On remote connect: the encrypted personal-data volume disconnects automatically.
Project team: Viktor Khudiakov (founder), Olga Khudiakova (design & brand), Alexey Skutin, Ph.D. (strategic advisor on architecture).
(1) Pre-seed/seed investment (2) SME customers for pilots (3) networking with those who can build on the architecture.