# Best open-source AI repos for self-hosted personal AI assistants (2026)

Source: https://dedupedai.com/best-ai-repos-for/personal-ai-assistants/ (Deduped, updated 2026-10-29). Markdown twin of the page; JSON: https://dedupedai.com/best-ai-repos-for/personal-ai-assistants/index.json

## What is the best open-source personal AI assistant you can run yourself?

The best open-source AI repos for self-hosted personal AI assistants right now are OpenClaw, Hermes Agent and AstrBot. Pick OpenClaw for one assistant on your own computer that you message from WhatsApp, Telegram, Slack, Signal or iMessage; Hermes Agent for a long-running agent on a server that learns reusable skills and remembers you; AstrBot when the assistant has to live in team chat apps such as Feishu, DingTalk, WeChat Work or Slack.

## Comparison

| Repo | What it does | Best for | Stars | Language |
|---|---|---|---|---|
| [OpenClaw](https://dedupedai.com/use-cases/openclaw-openclaw/) (openclaw/openclaw) | OpenClaw is an open-source personal AI assistant that runs on your own computer and answers you in the chat apps you already use, such as WhatsApp, Telegram, Slack, Discord, Signal and iMessage, plus native apps for desktop and phone. | People who want one always-on assistant they can message from their phone, running on hardware they control, and teams that want a shared deployment of the same thing. | 392,973 | TypeScript |
| [Hermes Agent](https://dedupedai.com/use-cases/nousresearch-hermes-agent/) (NousResearch/hermes-agent) | Hermes Agent is a self-improving AI agent from Nous Research. | Developers and power users who want a long-running agent that remembers them, learns reusable skills and works on a server while they message it from their phone. | 256,860 | Python |
| [AstrBot](https://dedupedai.com/use-cases/astrbotdevs-astrbot/) (AstrBotDevs/AstrBot) | AstrBot is an open-source, all-in-one agent chatbot platform that plugs LLMs into instant messaging apps such as QQ, WeChat Work, Feishu, DingTalk, Telegram and Slack. | Individuals, developers and teams who want an AI assistant, customer service bot or knowledge base inside the chat apps they already use. | 43,814 | Python |
| [OpenDots](https://dedupedai.com/use-cases/copilotkit-opendots/) (CopilotKit/OpenDots) | OpenDots is an open-source template from CopilotKit for persistent AI coworkers, each with its own computer, that move between web chat, mobile, Slack and calls. | Developers and startups who want to build their own always-on agent product rather than start from scratch. | 5,983 | TypeScript |

## When to pick which

### OpenClaw (openclaw/openclaw)

OpenClaw is an open-source personal AI assistant that runs on your own computer and answers you in the chat apps you already use, such as WhatsApp, Telegram, Slack, Discord, Signal and iMessage, plus native apps for desktop and phone. One Gateway process holds sessions, tools and channel connections; models and agent harnesses (Claude, Codex, local models) are swappable plugins. State, memory and credentials stay on your hardware, and there is no paid tier or hosted service.

Best for: People who want one always-on assistant they can message from their phone, running on hardware they control, and teams that want a shared deployment of the same thing.

Needs: macOS, Linux, Windows or WSL2 (the installer provisions Node.js if needed); An API key or plan for a model provider, or a local model; Accounts for the chat apps you want to connect.

Start here: [Run your own AI assistant on your laptop in 10 minutes](https://dedupedai.com/use-cases/openclaw-openclaw/run-your-own-ai-assistant-on-your-laptop/) (Beginner, 15 minutes)

1. On macOS, Linux or WSL2 run curl -fsSL https://openclaw.ai/install.sh | bash (Windows PowerShell: iwr -useb https://openclaw.ai/install.ps1 | iex)
2. If you already manage Node.js, use npm install -g openclaw@latest --allow-scripts=openclaw instead
3. Run openclaw onboard --install-daemon; onboarding verifies model access, creates the workspace and configures the Gateway
4. Check it with openclaw gateway status, then open the Control UI with openclaw dashboard

All 3 use cases: https://dedupedai.com/use-cases/openclaw-openclaw/ | GitHub: https://github.com/openclaw/openclaw

### Hermes Agent (NousResearch/hermes-agent)

Hermes Agent is a self-improving AI agent from Nous Research. It creates skills from experience, improves them while using them, keeps agent-curated memory, searches its own past conversations, and can run on a small VPS, a GPU box or serverless infrastructure. You talk to it from a full terminal UI or from Telegram, Discord, Slack, WhatsApp and Signal through one gateway, and it can run scheduled jobs in plain language. It works with Nous Portal, OpenRouter, OpenAI or your own endpoint.

Best for: Developers and power users who want a long-running agent that remembers them, learns reusable skills and works on a server while they message it from their phone.

Needs: macOS, Linux, WSL2 or Windows; An account or API key for a model provider (Nous Portal, OpenRouter, OpenAI or a custom endpoint); Optional: bot tokens for the messaging apps you connect.

Start here: [Install Hermes Agent and switch models without code](https://dedupedai.com/use-cases/nousresearch-hermes-agent/install-hermes-agent-and-pick-a-model/) (Beginner, 15 minutes)

1. Install on macOS, Linux or WSL2 with curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
2. Reload your shell, then start chatting with hermes
3. Choose your LLM provider and model with hermes model
4. Turn tools on or off with hermes tools, or run the whole wizard with hermes setup

All 3 use cases: https://dedupedai.com/use-cases/nousresearch-hermes-agent/ | GitHub: https://github.com/NousResearch/hermes-agent

### AstrBot (AstrBotDevs/AstrBot)

AstrBot is an open-source, all-in-one agent chatbot platform that plugs LLMs into instant messaging apps such as QQ, WeChat Work, Feishu, DingTalk, Telegram and Slack. It supports multimodal chat, agents, MCP, skills, a knowledge base, personas and auto context compression, integrates with Dify, Alibaba Cloud Bailian and Coze, offers 1000+ one-click plugins, an agent sandbox and a WebUI.

Best for: Individuals, developers and teams who want an AI assistant, customer service bot or knowledge base inside the chat apps they already use.

Needs: uv for the one-click install (or Docker); Python 3.12 via uv; An LLM provider and the bot credentials for each messaging platform you connect.

Start here: [Deploy an AI assistant into Telegram, Slack or Feishu](https://dedupedai.com/use-cases/astrbotdevs-astrbot/deploy-an-ai-bot-to-telegram-or-slack/) (Beginner, 30 minutes)

1. Run uv tool install astrbot --python 3.12
2. Initialize once with astrbot init, then start it with astrbot run
3. Open the WebUI and add your LLM provider
4. Connect a messaging platform such as Telegram or Slack with its bot credentials

All 3 use cases: https://dedupedai.com/use-cases/astrbotdevs-astrbot/ | GitHub: https://github.com/AstrBotDevs/AstrBot

### OpenDots (CopilotKit/OpenDots)

OpenDots is an open-source template from CopilotKit for persistent AI coworkers, each with its own computer, that move between web chat, mobile, Slack and calls. Agents can browse in an isolated browser, ask for human approval in chat, and keep working in the background. You host it yourself; conversations need CopilotKit Intelligence, either hosted, a local Docker evaluation, or a licensed self-hosted deployment.

Best for: Developers and startups who want to build their own always-on agent product rather than start from scratch.

Needs: Node.js 24 and npm; An OpenAI-compatible model key; CopilotKit Intelligence (hosted, local Docker evaluation, or licensed self-hosted).

Start here: [Run your own always-on AI coworker app from a template](https://dedupedai.com/use-cases/copilotkit-opendots/run-the-opendots-template-locally/) (Intermediate, 45 minutes)

1. git clone https://github.com/CopilotKit/OpenDots.git, then cd OpenDots and npm ci
2. Run cp .env.example .env and npm run dev, then open http://127.0.0.1:5173
3. For hosted Intelligence run npx copilotkit@latest login and npx copilotkit@latest project select
4. Add OPENAI_API_KEY and OPENAI_MODEL to .env and restart; do not run copilotkit onboard in this folder

All 3 use cases: https://dedupedai.com/use-cases/copilotkit-opendots/ | GitHub: https://github.com/CopilotKit/OpenDots

## Questions

### What is the best open-source personal AI assistant you can run yourself?

The best open-source AI repos for self-hosted personal AI assistants right now are OpenClaw, Hermes Agent and AstrBot. Pick OpenClaw for one assistant on your own computer that you message from WhatsApp, Telegram, Slack, Signal or iMessage; Hermes Agent for a long-running agent on a server that learns reusable skills and remembers you; AstrBot when the assistant has to live in team chat apps such as Feishu, DingTalk, WeChat Work or Slack.

### Which repo for self-hosted personal AI assistants is easiest to start with?

OpenClaw: its starter use case "Run your own AI assistant on your laptop in 10 minutes" is rated Beginner and takes about 15 minutes.

### How do I get started with OpenClaw?

On macOS, Linux or WSL2 run curl -fsSL https://openclaw.ai/install.sh | bash (Windows PowerShell: iwr -useb https://openclaw.ai/install.ps1 | iex); If you already manage Node.js, use npm install -g openclaw@latest --allow-scripts=openclaw instead; Run openclaw onboard --install-daemon; onboarding verifies model access, creates the workspace and configures the Gateway. Full steps and a copy-paste agent prompt: https://dedupedai.com/use-cases/openclaw-openclaw/run-your-own-ai-assistant-on-your-laptop/

### Are these self-hosted personal AI assistants repos free?

They are open-source projects on GitHub (check each license; some, like fair-code projects, limit commercial use). Some also sell a hosted version or need a paid model API key; the requirements for each repo are listed on this page.

### How is this list of self-hosted personal AI assistants repos chosen?

Deduped tracks the hottest AI repos on GitHub and writes step-by-step use cases from each README. For this page we picked the repos that do this job, ordered by fit and then GitHub stars. Repos with a risk flag or a retired page are left out. The page is rebuilt with the site; stars are as of the last build in 2026.

Third-party repos we don't own or vouch for. AI-written use cases can be wrong: review everything and run at your own risk.