EU AI Act Ready — Aug 2026

Your AI that learns you, not from you.

Self-hosted AI assistant that adapts to how you think, work, and communicate — while keeping every byte on your own servers.

By invitation only. We review every request personally.

Request received. We'll be in touch within 48 hours.
GDPR by architecture AES-256 encrypted Your data trains nothing
Privacy Router
Review my patient's treatment notes
Sensitive→ Local (on-device)
Explain flexbox vs grid layout
General→ Cloud (fast)
Check NDA clause for client García
Sensitive→ Local (on-device)
Draft a blog post about AI trends
General→ Cloud (fast)

KAI is a self-hosted, privacy-first AI assistant built by KAUFAST in Barcelona. It runs on your own server, keeps all data in an AES-256 encrypted database under your control, and uses smart privacy routing to automatically process sensitive queries (health, legal, financial) on-device while routing general queries to cloud AI for speed. KAI includes 13 productivity panels — tasks, reminders, notes, projects, meetings, email, calendar, habits, mood tracking, and more — in a single interface. It is GDPR compliant by architecture and designed for EU AI Act compliance (effective August 2026). Available on iOS, Android, macOS, and Linux.

13 panels
30 AI tools
AES-256 encrypted
iOS + Android native
<50MB total

Privacy Router

Sensitive data never leaves. Automatically.

Every message is scanned for sensitive content — health records, legal terms, financial data. Sensitive queries route to local inference. No toggle. No mode. It just works.

Sensitive query detected

“Review my patient's treatment notes from last Thursday and suggest a follow-up plan”
“Check the NDA clause about non-compete for client García”
● Local AI engine — zero data transfer
Same appSame UI

General query

“Explain the difference between flexbox and grid layout”
“Draft a blog post about AI trends in 2026”
● Cloud AI — frontier quality, auto-failover

How It Works

Deploy in minutes. Own your data forever.

KAI runs on any machine with Node.js. No GPU required. Under 50MB total.

01

Deploy to your server

Clone the repo, set environment variables, run node server/index.js. Works on a €7/month VPS, Raspberry Pi, or enterprise infrastructure.

02

Connect your services

Add Gmail, Proton Mail, Google Calendar, and your AI providers. KAI supports multi-provider fallback chains — fast US cloud, EU-native GDPR cloud, or fully local inference.

03

Use across all devices

Install the native app on iOS, Android, or macOS. Or use the PWA from any browser. Same encrypted data, same interface, same privacy guarantees.

04

KAI adapts to you

The more you use KAI, the more it learns your communication style, expertise, and preferences — stored locally in a database only you control.

K
KAI's Memory Vault
5 learned traits · encrypted
Tone preference
Direct and concise. Skip pleasantries. Prefers bullet-ready answers over long explanations.
Expertise level
Advanced developer — TypeScript, React, Node.js. No need to explain basics.
Personal context
Solo founder, three ventures, Barcelona. Multilingual: EN, ES, DE, CA.
Domain knowledge
SEO consultant pivoting to AI/GEO services. Understands Ahrefs, GSC, structured data.
Communication style
Copy-paste-ready commands over theoretical discussion. Execution first.

Adaptive Persona

KAI Remembers. KAI Adapts. KAI Is Yours.

Every conversation teaches KAI more about how you work. Your tone, expertise, clients, preferences — stored locally in a database you fully control.

Learns from every conversation

Vocabulary, preferences, recurring topics. After a week it feels like talking to a colleague who knows your business.

Per-conversation personas

Switch to “legal advisor” for contracts or “medical assistant” for patient processing. Each carries its own expertise.

Memory you control

View, edit, or delete anything KAI remembers. Export as JSON. Wipe with one click. Your hardware, your rules.

Client-specific knowledge

KAI remembers client names, case details, treatment protocols. Context that makes every interaction sharper.

Who It's For

Built for people who can't afford a data breach.

Designed for professionals where confidentiality isn't a preference — it's a legal requirement.

Healthcare

Patient summaries, treatment plans, referral letters. Sensitive data routes to local AI automatically — real isolation, not compliance theatre.

Legal

Contract review, case research, NDA drafting. Client names never reach the cloud. Ghost mode for privileged conversations.

Finance

Portfolio analysis, tax strategy, audit prep. AES-256 encrypted database with air-gap mode for fiduciary conversations.

Consulting

Client briefs, strategic notes, project personas. KAI adapts per engagement — different context, different expertise, one tool.

Executive

Board prep, investor updates, personnel decisions. Duress PIN for forced-unlock scenarios. Remote wipe for lost devices.

Privacy-First Individuals

Journaling, health tracking, personal planning. Your thoughts stay yours. No training data. No profiling. Full delete anytime.

Everything It Does

Built Lean. Ships Complete.

Under 50MB. Deploys with one command. No GPU required — plug in your model provider and go.

Streaming Chat Live
Real-time token streaming via WebSocket. Responses appear word by word. Markdown, code blocks, syntax highlighting built in.
Multi-Provider Routing Live
Fast US cloud, EU-native GDPR cloud, local on-device engine. Automatic fallback chain with config-driven provider switching.
Knowledge Base (RAG) Live
Upload books, manuals, documents. KAI chunks, embeds, and retrieves the most relevant passages per query automatically.
Voice Input & Output Live
Speak via browser microphone, transcribed by Whisper. Responses read aloud via local TTS — no audio data sent anywhere.
GDPR Toolkit Live
One-click data export (Article 20). Instant purge API (Article 17). Audit trails on every message. Consent banner included.
Encrypted Vault & Backup Live
AES-256-CBC encrypted backups with scrypt key derivation. Download a .enc file, store anywhere. Only your password decrypts it.
Gmail & Proton Mail Live
Read and compose via OAuth2 (Gmail) or Proton Bridge (localhost). All processing stays on your server.
Google Calendar Live
Two-way sync. Create, update, delete events. AI meeting briefs and recaps. Import events as meetings.
Installable PWA Live
Add to home screen on iPhone or Android. Offline-capable service worker. Instant updates, no app store review.
Push Notifications Live
Web Push via VAPID, APNs for iOS, FCM for Android. Reminder alerts and task deadlines, processed locally.
Self-Update System Live
Check for updates from within KAI. One-click pulls latest code, rebuilds, and restarts.
Smart Alarms Live
8 audio tones, snooze, in-app notification center with critical and informational tiers.
Habit & Mood Tracking Live
Daily habits with streaks. Mood logger with energy levels, tags, stats. Both feed into AI context.
Hardware-Grade Lockout Live
Escalating brute-force delays. Exponential backoff from 30s to 24-hour waits. Mathematically infeasible to guess.
Duress PIN & Remote Wipe Live
Emergency PIN shows fake empty state, silently exports encrypted backup and wipes data. Remote wipe via API.
AES-256 Database Encryption Live
SQLCipher encrypted at rest. Key derived via Argon2. Physical access to storage can't read your data.

Why It Matters

The cost of ignoring data privacy.

AI tools that store your data on third-party servers expose you to breach risk, regulatory fines, and loss of client trust. The numbers tell the story.

$4.88M
Average cost of a data breach in 2024
According to the IBM Cost of a Data Breach Report 2024, the global average cost reached $4.88 million — a 10% increase year-over-year and the highest total ever recorded.
€1.3B
GDPR fines issued in 2024 alone
The GDPR Enforcement Tracker reports cumulative fines exceeding €4.8 billion since 2018, with healthcare, legal, and financial sectors among the most targeted.
Aug 2026
EU AI Act becomes fully enforceable
The EU AI Act (Regulation 2024/1689) requires AI systems to meet transparency, data governance, and human oversight requirements. Self-hosted tools like KAI fall under the lowest risk category.

KAI eliminates these risks by keeping all data on your own server. No third-party storage means no third-party breach. GDPR compliance comes from architecture, not policy documents.

Efficiency & Sustainability

Local AI That Scales. Responsibly.

KAI's local processing engine runs on standard hardware — no GPU clusters, no cloud round-trips, no wasted energy. Built from day one to adopt compression advances as they ship.

Local Processing

Sensitive queries are processed entirely on your hardware. No data ever leaves your network. Runs on 8GB RAM — a €7/month VPS or a Raspberry Pi.

Zero egress No GPU needed 32K context
6x Context Expansion

Q2 2026: KV cache compression (peer-reviewed, Google Research) will extend local context to 100K+ tokens on the same hardware. No code changes — ships via the local inference runtime automatically.

TurboQuant Q2 2026 100K+ tokens
Lower Carbon Footprint

Local inference eliminates data centre round-trips. Quantised engines use a fraction of the energy of GPU clusters. KAI is designed to run efficiently on low-power hardware — privacy and sustainability, aligned.

~5W local inference No GPU clusters EU-hosted option

Efficiency Across the Market

KAI (Local)Cloud AI
Data leaves your networkNever (local mode)Every query
Hardware required8GB RAM, no GPUInternet + subscription
Inference cost€0 (your hardware)$20–100+/mo
Context window (current)32K tokens128K tokens
Context window (Q2 2026)100K+ tokens128K tokens
Energy per query~5W (local CPU)300W+ (GPU cluster)
Privacy guaranteeCryptographicPolicy-based

KAI is built from the start to adopt compression advances as they ship. When Google’s TurboQuant lands in the open-source ecosystem (Q2 2026), KAI’s local mode gets 6x more context — no update, no migration, no code change.

How KAI Compares

Productivity meets privacy.

No other tool matches KAI's combination of a full productivity suite, smart privacy routing, and enterprise security.

KAIChatGPTOpen WebUIJan.ai
Self-hosted
Smart privacy routing
Full productivity suite13 panelsChat onlyChat onlyChat only
Encrypted databaseAES-256
Duress PIN + remote wipe
Ghost mode
Air-gap operationPartial
GDPR by architecturePartialPartial
Native iOS + Android
Gmail + Proton MailBoth
Knowledge base (RAG)Plugins
Voice input + outputInput only

Available Everywhere

One App. Every Device.

Native on iOS, Android, and macOS. Installable as a PWA on Linux and any browser. Same experience, same security.

iPhone & iPad
App Store
Android
Google Play
macOS
Native App
Linux
Installable PWA

Questions & Answers

Everything You Need To Know.

Can't find an answer? Email us and we'll get back to you.

KAI is a privacy-first AI assistant that runs entirely on your own hardware. It combines the speed of cloud AI with the security of local processing — sensitive conversations never leave your network.

Professionals handling confidential data — lawyers, doctors, therapists, accountants, consultants. Privacy-conscious individuals and small teams.

Your data lives on your device. Sensitive queries are automatically kept local. And KAI is a full productivity suite — tasks, reminders, projects, notes, meetings, email, calendar — not just a chatbot.

Anything that runs Node.js. A €7/month VPS, your office server, or enterprise infrastructure. No GPU required.

Chat with AI, manage tasks, set reminders, organize projects with custom personas, brainstorm, take notes, manage meetings, track habits and mood, read emails, view calendar, feed books into a knowledge base — all from one interface.

KAI analyses every message for sensitive content. Sensitive topics are processed locally on your device. Everything else goes to a fast cloud provider with automatic failover. You always see which route handled your message.

Yes. KAI integrates via Proton Bridge. All email is decrypted and processed locally. No email content ever leaves your machine. Requires a paid Proton plan.

On your server. Period. Single encrypted database file on your disk. No third-party cloud databases, no analytics, no telemetry.

Yes — by architecture, not by policy. Data residency is wherever you deploy. Full Article 15, 17, 20 compliance. No data shared with any third party.

A secondary PIN that presents a clean empty state. Behind the scenes, it exports an encrypted backup and wipes the local database. For high-risk forced-unlock scenarios.

Yes. Full Local Mode routes every query to the on-device AI engine. Zero outbound connections. Fully air-gap capable.

Yes. Export as JSON, create encrypted backup, or delete everything with PIN confirmation. Ghost mode creates conversations that leave zero trace.

We deploy KAI for you on your own server or a private EU instance. Under 50MB total, no GPU required. We handle setup, updates, and maintenance.

KAI is currently available by invitation. Submit your email and we review every request personally.

Free tier includes full core features. Premium tiers add advanced security, managed hosting, and enterprise features. Currently by invitation only.

Yes. Private KAI instance on EU servers — SSL, custom domain, auto-updates, daily backups, email support included.

Your AI should know you. Not expose you.

KAI is currently available by invitation. Request access and we'll set you up.

Limited spots. We review every request personally.