Nanoswarm vs SSHush

Deploy personal AI agents that work for you 24/7 — chat with them in the app or connect them to Telegram. No servers, no API keys, no code. Nanoswarm lets you create, configure, and run your own AI-powered bots on dedicated cloud infrastructure. Give each one a personality, pick its tools, choose its model, and talk to it anywhere — in the built-in chat screen with streaming responses, or directly on Telegram. THREE ENGINES, ONE APP • Nanobot — Free. Fast, lightweight Python assistant with web search, file tools, and shell access. Great for daily tasks and quick automation. • OpenClaw (Pro) — Node.js agent with persistent memory, toggleable tools, and a rich skills community. Built for complex, long-running workflows. • Hermes (Pro) — Self-improving agent that writes its own code, delegates work to parallel sub-agents, and understands images and voice natively. Every engine runs in its own isolated container with dedicated storage. CHAT IN-APP OR ON TELEGRAM Talk to any bot directly from Nanoswarm's built-in chat — tap the Chat card on any running instance to start a conversation with streaming AI responses, message history, and abort controls. Prefer messaging? Link Telegram in seconds (no BotFather, no tokens) or bring your own bot token for full control. Your agent stays online 24/7 on both surfaces. CHOOSE YOUR AI MODEL Switch between Claude Opus 4.6, Claude Sonnet 4.6, Claude Haiku 4.5 (Anthropic), GPT-5.2 (OpenAI), and Kimi K2.5 (Moonshot). Swap models anytime. We handle provider routing so you never touch an API key. CREATE BOTS WITH PERSONALITY Pick a writing style — "No caps, no stress," "Crisp & polished," "Like texting a friend," or "Delightfully unhinged" — then pick a personality: personal cheerleader, straight-shooter advisor, high-energy operator, or deep-dive researcher. Add an emoji, custom lore, and instructions. Fine-tune anything later without recreating the bot. BUILT-IN TOOLS, NO CONFIGURATION • Web Search — live internet results • Web Fetch — read and extract any URL • File Operations — read, write, and edit workspace files • Shell — run commands in a sandboxed environment • Scheduler — cron jobs and recurring tasks • Sub-agents — background workers for parallel operations • Memory — persistent recall across conversations (OpenClaw & Hermes) • Vision & Voice — image and audio understanding (Hermes) • Browser automation and image generation available on supported engines OpenClaw lets you toggle individual tools on or off per bot. Every tool runs in an isolated cloud container — not on your phone, not on your laptop. RUN MULTIPLE BOTS FROM ONE DASHBOARD Create specialized agents for work, family, creative projects, or daily chores. Start, stop, restart, or delete any bot with a tap. Monitor live status, view container logs, and see message counts at a glance. PRIVACY AND SECURITY BY DEFAULT • Telegram tokens encrypted at rest (AES-256-GCM) • Each bot runs in its own isolated container • Server-side secrets never touch your device • Tenant isolation on every API call • Sign in with Apple, Google, or continue as a guest • Delete your account and all data anytime SUBSCRIPTION Nanoswarm Pro unlocks OpenClaw, Hermes, and premium AI models. Auto-renewable monthly subscription. Payment is charged to your Apple ID account at confirmation. The subscription renews unless cancelled at least 24 hours before the end of the current period. Manage and cancel in your Apple ID account settings. Download Nanoswarm — create your first bot in under a minute. Free to start, Pro available in-app. Terms of Use (EULA): https://nanoswarm.net/terms Privacy Policy: https://nanoswarm.net/privacy

3am. Your phone is face down in a drawer, on battery, in Low Power Mode. /var crosses 95%. You get woken up. That sentence is the entire product, and it is harder to deliver than it looks. An alert computed on your phone can only fire when iOS chooses to wake your phone, and iOS makes no guarantee that it ever will. Background execution is throttled on battery, suspended in Low Power Mode, stopped completely by a force quit, and starved hardest overnight - precisely when you need it. Worse, it fails in silence. You see your thresholds configured and you assume you are covered. TWO SYSTEMS, DELIBERATELY SEPARATE Live metrics come from your phone's own SSH session. Open the app, it connects, you see the server as it is right now. Alerting is a different system entirely. A small agent runs on your server and reports outward on its own schedule, independently of your phone. It does not care whether the app is open, backgrounded, force-quit or deleted from your home screen. A threshold breach originates on the server and arrives as a real push notification. If a server card shows as offline, that describes your phone's SSH connection. It says nothing about your alerting, which is still running. Your server speaks. Your phone listens. Nothing in between. Everything a monitoring platform does about a crossed threshold. None of the platform. No dashboard to host. No port to expose. No domain, no reverse proxy, no account. No platform sitting between your servers and you. No other iOS app does this. Every alternative either computes its alerts on the phone, which iOS can throttle or stop without telling you, or asks you to stand up and secure a dashboard of your own, or routes you through a monitoring platform with an account and a seat. FREE, AND NOT A TRIAL Unlimited servers. Full metrics. No agent, no account, no signup. Per-core CPU, temperature, memory, swap, disk by mount, load average, network interfaces. A full SSH terminal. Your servers sync across your devices through iCloud. This tier is agentless and always will be. Nothing is installed on anything. ALERTING, AS A SUBSCRIPTION Enable it on a server and the app installs the agent over the SSH connection you already have. There is nothing to sign up for and nothing to go and look at afterwards. Rules are configured in the app: CPU, memory, swap, disk per mount point, load average, temperature, interface down. And silence. If the agent stops reporting, you hear about it - the failure a monitoring system exists to catch, and the one a check on your phone can never see, because a sleeping phone and a dead server look identical from the inside. Every rule fires on sustained breach, not on a momentary spike. READ EVERY LINE BEFORE YOU INSTALL IT Alerting asks something real of you: a binary, on your server. You should not agree to that on trust. The agent is open source, Apache 2.0. github.com/ultra-nick/sshush-agent It runs as an unprivileged user, never as root, under systemd restrictions that make your filesystem read-only to it apart from one directory. It never listens on a port. It never executes a remote command. It never downloads anything, ever. Its whole network footprint is two URLs you can read in a config file. Metrics are sampled and rules evaluated on your server. Only a crossed threshold is ever sent. The numbers never leave the machine. Nothing else is touched. No package manager integration, no cron entry, no existing file modified. One command removes it completely, or the app asks it to remove itself. Your SSH credentials never leave your device. They live in your iCloud Keychain. Our infrastructure has never held an SSH credential and is not built to. Monitoring is free, unlimited and agentless. Alerting is there when silence stops being acceptable. --- Requirements: iOS 26.0 or later. iCloud syncs your servers and alert history across your devices. Terms of Use: https://www.apple.com/legal/internet-services/itunes/dev/stdeula/