LOCAL-FIRST AI AGENT OS

XUKUN_AI

YOUR AI. YOUR MACHINE.
One mind, many hands — an AI agent OS that lives on your computer, not in someone's cloud. She sees your screen and photos, writes code, runs your cameras, guards your network, automates your work, leads a team of AI agents that split big jobs and work them in parallel — and yes, she can even trade, every result published live, nothing hidden. Everything she knows lives in one local database you can hold in your hand.
GET XUKUN_AI ▸ ▶ WATCH HER WORK
● LOCAL-FIRST — YOUR DATA NEVER LEAVES HOME A TEAM OF AGENTS, ONE DOOR SHE CODES & SEES SENTINEL EYES — CAMERAS WORKS OFFLINE TOO macOS + LINUX + ANDROID ₿ BTC · LIGHTNING · XMR · USDT
Don't trust the AI. Verify her. — her trading desk publishes a live verifiable record ▸: losses included, decisions hashed, settling on-chain. P&L … · … decisions hashed · … won/lost
XUKUN_AI — the world view, live on a real machine
The XukunAI world view — Athena and her crew, running locally
▸ THIS IS THE PRODUCT — not a mockup. Athena steering her crew, rendered live by the engine itself.
// WATCH IT
The machine, in motion
This demo was voiced, assembled and verified by XukunAI's own studio — the same engine your agents use. No editor, no stock footage house: the pictures are the product, the voice is the product, the video is the product.
// HOW IT WORKS
Running in three minutes
No cloud account, no console to babysit. Three steps and your crew is alive.
01
INSTALL
One-click installer for macOS or Linux. Double-click, done — the app lives entirely on your machine.
02
CONNECT A BRAIN
Bring your own AI key — Claude, GPT, Grok — or flip on the built-in local brain and pay nothing per token. Your prompts never pass through us either way.
03
MEET YOUR ARCHITECT
Athena wakes up and takes the wheel — she multitasks, steers her own crew of specialists, and shows their work live. You just talk to her; she handles the rest.
// CAPABILITIES
An operating system for AI agents
One Architect leads a team of specialist agents. You watch them think, delegate and work — live, as glowing particle orbs on your desktop.
◈
ONE MIND, MANY HANDS
You talk to ONE agent — your Architect. She multitasks, steers her crew of specialists, and shows their work as live indicators that fade when done. Queue your messages while she works; steer her mid-run. Never ten chat windows — one conversation, many hands.
👁
SHE SEES
Drop a photo in chat — a room, a screenshot, a document — and the local vision brain reads it on your machine. Ask her to restyle your room, explain an error dialog, or summarize a scanned page. No cloud image API; the pixels never leave home.
◇
SHE CODES
A real coding agent rides on your machine. Say 'build a website and put it in this folder' — it lands exactly there, and she verifies the result herself before reporting done. Auto-installed, sandboxed, killable with one word.
⬡
THE FLEET MESHES
Your machines work as one. 'Run this on the studio Mac' from any chat — the task crosses the mesh, executes there, and the result lands back in the same conversation. Every machine sees the fleet's health on one board.
✦
KNOWLEDGE CONSTELLATION
Every document your agents read or write settles into a living 3D star map. Related knowledge lines up across folders — you can literally see what your AI knows. And it works silently: knowledge grounds her answers without ever cluttering the chat.
✈
THE PHONE, FOR REAL
The free XUKUN Mobile app is the same one-mind chat in your pocket — attach photos straight from the camera, gallery or files, watch the crew's work chips, steer mid-run. Over your Wi-Fi or worldwide through our own relay — your Mac stays home, your agents come with you.
◷
AUTOMATIONS
Schedule daily briefings, weekly reports, monthly digests — or trigger agents from the outside world via webhooks. They run even while you sleep.
▣
CONNECTS TO ANYTHING
Six integration worlds — Security Ops (100+ tools), Business, Home, Dev, Communication, AI. Search never fails: name any service and your Architect builds the bridge, tests the key, locks it in the local vault.
⛁
TRAINABLE SKILLS
Teach your agents your procedures — upload playbooks, link integrations, assign skills per agent. Your workflow becomes their instinct.
// USE CASES
What will your team run?
Real workflows XukunAI operators build in their first week — each one is an agent, a skill and an automation working together.
⚙
BUSINESS AUTOMATION
An ops agent that drafts invoices from your notes, chases overdue payments on schedule, summarizes your CRM pipeline every Monday 08:00 and files every report into your private knowledge base — no Zapier tax, no data leaving your machine.
⛨
PENTEST & RED TEAM
A recon agent trained on your methodology: scope in, OSINT sweep out. Findings settle into the constellation per client, a report-writer agent drafts in your template, and nothing about the engagement ever touches a third-party SaaS.
◎
THREAT INTEL & SOC
Bots watch feeds and webhooks around the clock; an intel agent correlates against your own past cases — a memory no cloud copilot has — and briefs you on your phone before your coffee. Escalation stays a human decision.
⛁
MSP / CONSULTANCY
One agent per client, walls between them. Weekly status digests write themselves from each client's knowledge folder, and license tiers scale from a solo operator to a ten-agent bench for the whole practice.
✍
RESEARCH & CONTENT
A researcher digs, a writer drafts, the Architect merges — credited teamwork on every deliverable. Every source they touch becomes a permanent, searchable star on your knowledge map.
⌂
HOME COMMAND CENTER
Morning briefing at 07:00, family documents in a private vault, smart-home hooks — and your security cameras watched by SENTINEL EYES, fire-alert escalation on the roadmap. Your household's AI, on your hardware, answerable to no one else.
// SEE IT
This is what your machine looks like with a crew on it
Real screenshots, zero mockups — the actual product running on a Mac.
XukunAI one-door chat — Athena with live work rail and crew trace
ONE DOOR — one conversation with Athena; her work rail shows the crew's steps live, and the crew trace rides quietly under each answer
XukunAI WORLD view — living agent orbs
WORLD — your agents live here: particle entities, live activity, briefings and calendar at a glance
Knowledge Constellation — 3D star map of your AI's knowledge
KNOWLEDGE CONSTELLATION — every doc becomes a star; drag, pan, zoom what your AI knows
SENTINEL EYES — security cameras with AI event descriptions
SENTINEL EYES — your cameras on watch: motion caught, snapshot taken, and the AI writes down what it saw
// SENTINEL EYES
Cameras that explain what they saw
Plug in the security cameras you already own — XukunAI turns them from dumb recorders into a crew member that watches, understands and reports.

👁 YOUR CAMERAS, ON WATCH

Hikvision, Dahua, Axis, Reolink, Amcrest — one snapshot URL and the eye is live.
  • Motion detection on-device — no cloud subscription, no per-camera fees
  • AI describes every event — "a person in a dark jacket at the side door" in your journal, not a folder of JPEGs
  • Alerts reach your pocket — through the XukunAI mobile bridge, nothing third-party
  • Ask your agents — "what did the cameras see today?" answered from the real event log
LIVE SINCE 3.53

⌂ LOCAL EYES

Pick a vision-capable local model and every frame is analyzed on your own machine. Footage of your home never touches a cloud API — private, free, and exactly as smart as the Mac you run it on.
[EYES] frame described locally — never left this machine
LIGHT · 1 EYE GUARD · 3 EYES HEAVY · 10 EYES FIRE-ALERT ESCALATION · ROADMAP
// THE HIVE ● LIVE — ONE MIND ACROSS YOUR MACHINES
Every machine makes every machine smarter.
Your Local Brain lives on your machine, learns your world and answers to you alone. The Hive is what the fleet shares: HIVE-MIND SYNC keeps every machine's knowledge in step — what any of your machines learns, every machine knows, synced nightly and on demand. And the Xukun Brain distills hard-won know-how into skill packs shipped to every licensed machine — your agents download new abilities while you sleep. Your conversations, documents and memories never make the trip — skills travel, secrets don't.
⌂
YOUR LOCAL BRAIN
Reads your documents, remembers your preferences, masters your routines — and keeps it all in one database on your own disk. What this brain knows is between you and your agents. Nobody else. Not even us.
⬡
THE XUKUN BRAIN
Our learning core on the mothership. It collects lessons — never data — from the whole fleet, refines them, and ships the upgrades back down. One agent learns to spot a new scam; every agent develops the immunity. How it learns is our own technology — and it stays our secret.
TLS
SKILLS TRAVEL. SECRETS DON'T.
Between the two brains runs a one-way rule: know-how may rise, your life may not. Chats, documents, memories, keys — there is simply no road from your private data to the network. Sharing is opt-in, always.
This is not a promise — it is running. When a task matches know-how your machine doesn't have yet, the right pack installs itself mid-task and your agents use it immediately: zero extra API cost, working even offline. Collective intelligence, private execution. No other AI company can say that sentence.
// TWO WORLDS. ONE OS.
Built for professionals and for home

⛨ SECURITY OPERATORS

Pentesters, analysts, MSPs — a private workbench that never leaks client data to a SaaS.
  • Recon & threat-intel automations
  • Offensive / defensive tool integrations
  • Report-writing agents trained on your templates
  • Local brain built in — one click installs it, camera vision included

⌂ EVERYDAY LIFE

A calm, beautiful AI home base for mail, plans, research and the smart home.
  • Morning briefings before you wake
  • Email & calendar integrations
  • Family knowledge base that stays home
  • Camera alerts, straight from your own CCTV
// THE NUMBERS — LIVE
She grades herself in public
Every figure below comes live from the product the moment you load this page — not marketing copy. Each decision is cryptographically fingerprinted so the record can't be quietly rewritten, and live bets settle on-chain from a public wallet anyone can inspect.
…
SETTLED P&L · USD
…
DECISIONS HASHED
…
DISCIPLINE · PASS RATE
…
ARMOR · ORDERS VETTED 7D
OPEN THE FULL LIVE RECORD ▸
// ABOUT US
Built in Dubai
XUKUN_AI is built by Kyber IT Solutions, an AI company headquartered in Dubai, United Arab Emirates. We build assistants that live on your own hardware instead of our cloud: your conversations, files and memories stay in one database on your machine, and every claim this product makes about itself — right down to its trading record — can be checked live on this site.
COMPANY
Kyber IT Solutions
Mother company of XUKUN_AI
HEADQUARTERS
Dubai, UAE
AI built in the Emirates
WHAT WE BELIEVE
Your AI. Your machine.
Private by architecture, honest by design
CONTACT
support@xukunai.com
A human reads it
// THE TECH — IN PLAIN WORDS
What's actually inside
No black box and no buzzwords — this is the exact stack you install.
RUNS ON
macOS and Linux desktops, plus an Android app. Installs in minutes; updates ship signed from this site and the app tells you when one is ready.
YOUR DATA
Everything she knows lives in one local database file on your disk — chats, memories, documents, settings. We run no data cloud. Delete the file and she forgets; copy it and you've moved her.
THE BRAINS
Bring your own AI: she works with 8 providers (OpenAI, Kimi, DeepSeek, Grok, Gemini, Mistral, OpenRouter — or fully local). No internet or no keys? She measures your computer and runs the strongest local model it can handle, so she never goes dark.
THE TEAM
One lead agent — Athena — is your single door. Under her you create agents and bots, each with a role. Give her a big job and she splits it across the team, they work in parallel, and she hands you back one merged answer. Add or permanently delete team members with one click.
TRADING
One capability among many — and the one she proves in public. Paper-trading by default; real money only after you arm it, and only after a strategy passes three separate statistical exams on real market data. The complete live record — wins, losses, how honest her predictions were — is published on the proof page: an example of the transparency every capability inherits.
SECURITY
Admin controls answer only on your machine, API keys live in an on-device vault and never leave it, a network guard watches your LAN, and there's a visible stop switch. Licenses are per-machine.
PAYMENTS
Card checkout, or crypto: Bitcoin, Lightning, USDT and Monero — processed on our own payment server, not a third-party gateway.
// PRICING
One license. One year. No data hostage.
Bring your own AI key (Anthropic / OpenAI) — you pay the model provider directly, we never see your traffic. No per-seat traps, no upsell popups. And the ONE MIND brain economy stretches every token: your premium brain plans, free and local brains do the typing.
LIGHT
€299 / YEAR
≈ €25 / month
  • 1 Architect + 3 agents
  • 1 camera eye — SENTINEL EYES
  • Automations & webhooks
  • Knowledge constellation
  • Six integration worlds — connect anything
  • XUKUN Mobile bridge — agents in your pocket
  • 1 year of updates
BUY LIGHT ▸
MOST POPULAR
GUARD
€499 / YEAR
≈ €41 / month
  • 1 Architect + 6 agents
  • Up to 10 automation bots
  • 3 camera eyes — SENTINEL EYES
  • Everything in LIGHT
  • Room to grow the hive
  • 1 year of updates
BUY GUARD ▸
14-DAY MONEY-BACK
HEAVY
€799 / YEAR
≈ €66 / month
  • 1 Architect + 10 agents
  • Up to 20 automation bots
  • 10 camera eyes — SENTINEL EYES
  • Everything in LIGHT
  • COUNCIL — on hard questions, every brain you own deliberates and votes; disagreements surface to you instead of one model guessing alone
  • Trainable skill packs
  • Priority support
  • 1 year of updates
BUY HEAVY ▸
WE ACCEPT: CARDS ₿ BITCOIN ɱ MONERO — SOON — because privacy shouldn't stop at checkout.
🛡 HEAVY tier: 14-day money-back guarantee — if it's not for you, email support within 14 days for a full refund. No questions, no hoops.
// DOWNLOAD
Already licensed? Get the app.
Enter your license key — we verify it and hand you the latest build. One key covers macOS and Linux (Windows in development). Mac users get a one-click installer — double-click, done. Your key arrives by email after purchase.
Free companion app — your agents in your pocket, at home and anywhere through the XUKUN RELAY. Direct APK, no Play Store — on purpose: no store account, no Google tracking, no third party between you and your own machines. Requires XukunAI running on your computer. iOS is not planned yet — Android first, done right.
DOWNLOAD FOR ANDROID ⬇
Android will ask permission to install apps from your browser — that's normal for any app outside a store. Tap Allow, then Install.
XukunAI is Unix-native today — the security tooling, local brain and camera stack are built for macOS and Linux. The Windows port is in active development. Leave your email and we'll send one message when it ships. Nothing else, ever.
// SHIPPED LATELY
A product that moves every week
Real releases, real changelogs — this is what active development looks like.
RECENT RELEASES ▾
  • v4.3.6 HEAR, ACT, REMEMBER — she reflects your request back before acting on real work (never builds the wrong thing), never narrates process instead of delivering, and weaves your shared history in naturally. INTENT-LOCK: she builds what you NAMED, not her own template
  • v4.3.0 HIVE-MIND SYNC — the fleet shares one mind: what any machine learns, every machine knows. Plus GUARDIAN-TRUST (mark your device herd in one tap — the guard stops crying wolf on YOUR devices) and the OBSIDIAN lane (your vault becomes her knowledge, locally)
  • v4.2.0 MESH-ADDRESS + MEMORY-FRESHNESS — the fleet is dialable worldwide through the relay, and old facts carry their age so she hedges instead of asserting stale truth
  • v4.0.0 CALM — one door only: the cockpit rebuilt cleaner, faster, brighter. Dead code killed, polls calmed, zero performance cost
  • v3.131 SILENT PORTAL + KIMI TARGET LAW — knowledge from your linked worlds grounds her answers invisibly, and 'build X, put it in this folder' lands exactly there — verified by her own eyes
  • v3.129 ONE DOOR + EYES + AURORA — Athena is the only conversation: she multitasks and steers the crew while you watch. Queue your messages, steer her mid-run. She sees your photos now. The whole cockpit rebuilt fresher and brighter
  • v3.124 MESH TASKS + KIMI CODE LANE — ask any machine to work from one chat; a real coding agent on your machine, auto-installed, Athena-verified
  • v3.120 MINIMIZE LAW — the chat collapses to a clean line when you want focus, one click brings the crew back
  • v3.119 CORTEX SURFACE — one mind, visible: Athena's tab is the front door, and every answer shows the slim crew trace of who worked on it
  • v3.117 VOICE WARMTH + MULTI-ATTACH — talk naturally, drop several files at once, size the chat to your screen
  • v3.115 HIGGSFIELD LANE — real AI-generated footage in the studio, rendered scene by scene on your machine
  • v3.113 FLEET SIGHT — your machines see each other; health, versions and work visible on one board
  • v3.111 WORK-TREE LAW — big jobs split into a visible tree of work; nothing dies silently, everything resumes
  • v3.108 PROMO-CRAFT — the studio learned pro video craft: kinetic type, device-framed product shots, hook-first stories
  • v3.104 SUPERHUMAN PROTOCOL — she notices unfinished work, remembers where you left off, and acts
  • v3.97 TRUE TOKEN STREAMING — answers are born live on your screen, not after a wait
  • v3.92 CACHE LAW — the built-in webview can never show you a stale page again
  • …and 28 releases in between — the changelog never sleeps: real fixes, real laws, every week
// FAQ
Questions
Where does my data live?

In a single SQLite database on your own machine. Conversations, memories, documents, automations — all local. Export everything with one click, delete it whenever you want. The mothership only knows your license key and email.

Which AI models does it use?

You bring your own API key and switch brains in Settings: Anthropic Claude, OpenAI GPT, xAI Grok — or run 100% offline with a local Ollama model. Your prompts go directly from your machine to the model provider (or never leave it at all) — never through our servers.

What do I need to run it?

A Mac (Apple Silicon or Intel) or a Linux machine, plus an AI API key — or none at all if you run the local Ollama brain. Windows is in development.

How do crypto payments work?

Pick crypto at checkout and you land on our self-hosted BTCPay Server — scan the QR, send the exact Bitcoin amount, and your license key is emailed after confirmation. No payment processor, no middleman: the invoice comes from our own node. Monero support is coming.

Is there a money-back guarantee?

On the HEAVY tier (€799/year), yes — 14 days, no questions asked. Email support within 14 days of purchase for a full refund. LIGHT and GUARD are final-sale to keep prices honest.

What is the Hive?

It's live, on two levels. Your Local Brain lives on your machine and learns your world. The Xukun Brain — our learning core — distills what agents everywhere learn into skill packs and ships them to every license: 63 packs so far, installed automatically the moment your work needs them. And since V4.3 the fleet itself is a hive mind: HIVE-MIND SYNC keeps every machine's knowledge in step — what any of your machines learns, every machine knows, synced nightly across your own hardware. Only know-how travels; your conversations, documents and memories never leave your machines. How the Xukun Brain learns is our own technology.

What happens when my year ends?

The app keeps working. Renewal extends updates, new integrations and support for another year — your data is yours either way.

GET XUKUN_AI

CARD
₿ CRYPTO