Hermes Agent Setup: Free AI Assistant on Telegram (5-Min Guide)
Last updated: 17 July 2026
Let's be clear about something upfront — most AI agent tutorials are vague. They tell you agents are "transforming finance" but never show you the actual terminal commands to get one running. Today, we're fixing that.
Hermes Agent, built by Nous Research, is an open-source CLI AI assistant that runs on your own machine, connects to your preferred LLM provider (OpenAI, OpenRouter, or local models), and exposes itself as a chat interface through Telegram. That means you can interact with your agent from your phone while it operates from your laptop or a VPS — perfect for finance professionals in Mumbai, Bengaluru, or Delhi who want automated research and portfolio monitoring on the go.
In this tutorial, you'll go from zero to a running Hermes Agent that responds to Telegram messages, uses an OpenAI API key for model inference, and persists memory across sessions. Every step is exact — no hand-waving.
What You'll Need Before Starting
- A machine running Linux/macOS/WSL2: Hermes Agent is a Python-based CLI. It runs natively on Linux and macOS. On Windows, use WSL2 with Ubuntu.
- Python 3.10+ installed and accessible from your terminal.
- An OpenAI API key: Generate one at platform.openai.com/api-keys. You'll need credits on your account — for Indian developers paying in INR, expect roughly INR 2,000 to 3,000 to load into your account. Budget-conscious? Skip to the OpenRouter section below which is often 3-5x cheaper.
- A Telegram Bot Token: You'll create this in under 2 minutes using BotFather.
- Git for cloning repositories.
Step 1 — Clone and Install Hermes Agent
Hermes Agent can be installed via the official install script or pip. For developers, the pip route gives you more control.
# Option A: Official install script (recommended)
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
# Option B: Pip install
pip install hermes-agent
# Verify installation
hermes --version
The install script handles Python dependencies, creates the config directory at ~/.hermes, and sets up the initial configuration. You'll see a welcome message confirming the install.
Step 2 — Run the Setup Wizard
The setup wizard walks you through provider selection, model choice, and messaging platform configuration:
# Start the interactive setup
hermes setup
# Or use the portal for a simpler experience
hermes setup --portal
The wizard will ask you to:
- Choose a provider: Select OpenAI for this tutorial (you can add OpenRouter later for cheaper models)
- Enter your API key: Paste the OpenAI API key you generated earlier
- Select a model: GPT-4o is the default, but GPT-4o-mini is cheaper for testing
- Choose messaging platform: Select Telegram when prompted
Step 3 — Create Your Telegram Bot
Open Telegram and search for @BotFather — Telegram's official bot for creating bots.
- Send
/newbot to BotFather
- Choose a name for your bot (e.g., "My Finance Agent")
- Choose a username (e.g., "my_finance_agent_bot")
- BotFather will reply with your bot token — copy it immediately
The token looks like: 1234567890:ABCdefGHIjklMNOpqrSTUvwxYZ
Step 4 — Configure Hermes Agent with Telegram
Now connect your bot to Hermes Agent:
# Start the messaging setup
hermes setup messaging
# When prompted:
# 1. Select "Telegram" as the platform
# 2. Paste your bot token from BotFather
# 3. Enter your Telegram user ID (get it from @userinfobot)
# 4. Confirm the configuration
To get your Telegram user ID, search for @userinfobot on Telegram and send /start. It will reply with your numeric user ID.
Step 5 — Start the Gateway
The gateway is the background process that connects Hermes to Telegram:
# Start the gateway in the background
hermes gateway start
# Check status
hermes gateway status
# View logs
hermes gateway logs
Once the gateway is running, open Telegram, find your bot, and send it a message. Hermes Agent will respond using your OpenAI API key.
Step 6 — First Conversation
Try these finance-focused prompts to test your agent:
# Basic test
"Hello, what can you help me with?"
# Finance-specific
"Summarize today's Nifty 50 performance"
# Research task
"Search for the latest SEBI regulations on mutual fund commissions"
# Analysis
"Compare returns of HDFC Flexi Cap vs ICICI Prudential Flexi Cap over 3 years"
Your agent will use OpenAI's GPT model for reasoning, and can search the web, read files, and execute code — all through the Telegram interface on your phone.
Step 7 — Adding Skills for Finance Workflows
The Hermes Agent skill system lets you create reusable workflows for finance tasks:
# Create a skill for daily market summary
hermes skill create market-summary
# The skill file will be created at:
# ~/.hermes/skills/market-summary/SKILL.md
Skills are procedural memory — the agent creates them from experience and reuses them. For finance professionals, useful skills include:
- Daily market summary: Fetch Nifty and Sensex data, analyze top gainers and losers
- Portfolio rebalancing: Check allocation against your target percentages
- Tax planning reminders: Alert about 80C and 80CCD investment deadlines
- SIP tracking: Monitor monthly SIP deductions and NAV updates
Step 8 — Setting Up Cron Jobs for Finance Automation
Hermes Agent can run tasks on a schedule — perfect for finance automation:
# Create a daily morning briefing at 9am
hermes cron create --name morning-briefing --schedule "0 9 * * *" \
--prompt "Send me today's market summary with Nifty, Sensex, and top 5 gainers"
# Create weekly portfolio check every Monday at 10am
hermes cron create --name portfolio-check --schedule "0 10 * * 1" \
--prompt "Check my portfolio allocation and suggest rebalancing if needed"
Cost Comparison: Hermes Agent vs Alternatives
| Feature | Hermes Agent | ChatGPT Plus | Claude Pro |
| Monthly cost | INR 0 to 3,000 (API usage) | INR 1,700 | INR 1,700 |
| Telegram integration | Native support | Not available | Not available |
| Memory persistence | Across all sessions | Limited context | Limited context |
| Skill creation | Agent creates skills | Not available | Not available |
| Cron jobs | Built-in scheduler | Not available | Not available |
| Browser control | Full web control | Not available | Not available |
| Self-hosted | Your machine or VPS | Cloud only | Cloud only |
Troubleshooting Common Issues
| Issue | Cause | Solution |
| Telegram bot not responding | Gateway not running | Run hermes gateway restart |
| API key error | Invalid or expired key | Check key at provider dashboard |
| Memory not persisting | Session not saved | Check ~/.hermes directory permissions |
| Cron job not running | Gateway stopped | Restart gateway, check cron list |
| Model not responding | Provider rate limit | Switch model or wait for reset |
Security Best Practices
- Never share your Telegram bot token: Anyone with the token can control your bot
- Use allowlists: Only allow your Telegram user ID to interact with the bot
- Enable command approval: Require approval for dangerous commands
- Run in container: Use Docker for isolation from your main system
- Monitor API usage: Set spending limits on your API provider dashboard
Next Steps from PyBankers
Sources
- Hermes Agent — Quickstart Guide — checked 17 July 2026, confirmed installation, provider setup, Telegram integration
- GitHub — NousResearch/hermes-agent Repository — checked 17 July 2026, confirmed 27K+ stars, MIT license, feature list
- Heyuan110 — Hermes Agent Review 2026 — checked 17 July 2026, confirmed v0.9.0 features, 5 USD VPS setup
- DataCamp — Hermes Agent Setup and Tutorial Guide — checked 17 July 2026, confirmed installation, skills, memory system
- Hermes Agent — Telegram Messaging Documentation — checked 17 July 2026, confirmed BotFather setup, topic mode, file handling
Last updated: 17 July 2026. Hermes Agent is open-source and updated frequently. Always check the official documentation for the latest installation steps. API costs vary by provider and usage. Financial advice generated by AI agents should be verified with a qualified financial advisor before making investment decisions.
The VPS Route: Running Hermes 24/7
For finance professionals who need their agent running around the clock, a VPS is the way to go. Here's the quick setup:
# Deploy on a VPS (e.g., Hostinger, Hetzner, DigitalOcean)
# SSH into your VPS
ssh root@your-vps-ip
# Install Docker
curl -fsSL https://get.docker.com | sh
# Pull and run Hermes Agent
docker pull ghcr.io/nousresearch/hermes-agent:latest
docker run -d --name hermes \
-v ~/.hermes:/root/.hermes \
-p 3000:3000 \
ghcr.io/nousresearch/hermes-agent:latest
A $5/month VPS from Hetzner or Hostinger is plenty for running the gateway. This gives you 24/7 availability — your agent responds to Telegram messages even when your laptop is closed.
OpenRouter: The Budget-Friendly Alternative
If OpenAI API costs are too high, OpenRouter gives you access to dozens of models at lower prices:
| Model | Provider | Input Cost (per 1M tokens) | Output Cost (per 1M tokens) |
| GPT-4o | OpenAI | $2.50 | $10.00 |
| GPT-4o-mini | OpenAI | $0.15 | $0.60 |
| Claude 3.5 Sonnet | Anthropic | $3.00 | $15.00 |
| DeepSeek V3.2 | DeepSeek | $0.28 | $0.42 |
| Llama 3.1 70B | Meta | $0.52 | $0.75 |
For Indian developers on a budget: DeepSeek V3.2 at $0.28/$0.42 per million tokens is 14x cheaper than GPT-5.6 Luna. That means your INR 2,000 monthly budget goes much further.
Model Selection Strategy for Finance
Not all models are equal for finance tasks. Here's what works best:
- GPT-4o: Best for complex financial analysis, multi-step reasoning
- GPT-4o-mini: Good for daily summaries, quick questions, budget-friendly
- Claude 3.5 Sonnet: Excellent for document analysis, long-form reports
- DeepSeek V3.2: Great value for routine finance tasks
- Hermes 4.3: Fine-tuned for agent tasks, good at following instructions
Pro tip: Use GPT-4o for complex analysis and GPT-4o-mini for routine tasks. This hybrid approach cuts costs by 60% while maintaining quality where it matters.
Indian Finance Use Cases
Here are real-world finance workflows you can automate with Hermes Agent:
- Morning market briefing: Daily Nifty/Sensex summary with sector analysis delivered to Telegram at 9am
- SIP tracker: Monthly reminder of SIP deductions, NAV updates, and portfolio performance
- Tax planning assistant: Quarterly reminders for 80C/80CCD investments with personalized suggestions
- News aggregator: Curated financial news from Economic Times, Moneycontrol, and Livemint
- Budget analyzer: Monthly expense categorization from bank statement CSV exports
Advanced: Multi-Model Routing
For power users, Hermes Agent supports model routing — using different models for different tasks:
# In your config.yaml, set up model routing
model_routing:
# Use cheap model for simple tasks
simple_tasks: "openai/gpt-4o-mini"
# Use powerful model for complex analysis
complex_tasks: "openai/gpt-4o"
# Use local model for privacy-sensitive tasks
private_tasks: "ollama/llama3.1:8b"
This way, a simple "What is the Nifty at?" uses GPT-4o-mini (cheap), while "Analyze my portfolio and suggest rebalancing" uses GPT-4o (powerful).
📌 Editorial Disclosure & Verification
Researched and drafted with PyBankers AI research pipelines; fact-checked, edited, and verified by the PyBankers editorial team. This report is for educational and informational purposes only and does not constitute financial, legal, or investment advice. Verify critical details independently.