Run Hermes Agent, the self-improving AI agent from Nous Research, from a pre-configured image in your own cloud account. It remembers past conversations, builds its own skills and answers on Telegram, Discord, Slack, WhatsApp and Signal, with the LLM provider of your choice.
Hermes Agent is an MIT-licensed, self-hosted autonomous AI agent built by Nous Research. It keeps memory across sessions, learns new skills from experience and runs scheduled automations. One gateway connects it to your messaging platforms, and it works with any LLM provider rather than a single vendor's model.
Remembers past conversations using full-text search and LLM summaries.
Learns from experience and keeps the skills it builds.
Telegram, Discord, Slack, WhatsApp and Signal from one deployment.
Run recurring jobs with the built-in cron scheduler.
Extend the agent with plugins and Model Context Protocol servers.
Nous Portal, OpenAI, OpenRouter, Anthropic, Vertex AI, Bedrock or custom endpoints.
Running Hermes Agent yourself means runtimes, a gateway process, a reverse proxy, SSL and backups. The marketplace image ships with that done, plus a web dashboard.
Subscribe on AWS Marketplace or Google Cloud Marketplace, enter your domain and admin username and password, and deploy.
Create a DNS A record for your Hermes domain with the server's public IP, then sign in to the dashboard once SSL has issued.
SSH into the server and run hermes model inside the container to pick a provider and add your API key.
Open Channels in the dashboard and pair Telegram, Discord, Slack, WhatsApp or Signal.
An AMI on AWS Marketplace, launched into your own AWS account with a CloudFormation stack.
Launch on AWS 
The same agent as a Google Cloud Marketplace image, deployed into your own GCP project.
Launch on Google Cloud | Hermes Agent by Meetrix | Hosted AI agent SaaS | Self-managed install | |
|---|---|---|---|
| Licence | MIT, open source | Proprietary | MIT, open source |
| Pricing model | Instance and software fee, plus your own model cost with no markup | Per-seat or usage markup on model cost | Server cost plus your own model cost |
| Platforms | Telegram, Discord, Slack, WhatsApp and Signal from one gateway | Usually one, more as paid add-ons | Whatever you wire up yourself |
| Data location | Your own AWS account or GCP project | Vendor's infrastructure | Your server |
Model tokens are billed by your provider directly, not by Meetrix. Read about Hermes Agent on AWS →
A self-hosted, self-improving autonomous AI agent built by Nous Research. It keeps memory across sessions, builds its own skills, runs scheduled automations and connects to Telegram, Discord, Slack, WhatsApp and Signal through one gateway.
Yes. The image does not include model access. Run hermes model after deployment and choose Nous Portal, OpenAI, OpenRouter, Anthropic, Google Vertex AI, AWS Bedrock or a custom OpenAI-compatible endpoint. The provider bills you directly.
Yes. Memory persists using SQLite FTS5 full-text search combined with LLM summarisation, stored on the server's own disk, so it survives a stop and restart.
Open the Channels section of the dashboard and follow the pairing flow for each platform. They all run through the same gateway.
The username and password you set at deployment. There is no default credential. To change them later, regenerate /etc/nginx/.htpasswd on the server and reload nginx.
t3a.small on AWS, or the default General purpose type on Google Cloud. Size up for many concurrent automations or several busy channels.
On the server in your own AWS account or GCP project. Conversation history, the memory database and your API keys stay there; Meetrix has no access to the running deployment.
Make sure DNS has propagated to the server, then SSH in and run sudo bash /root/certificate_generate_standalone.sh.
Your own agent, without someone else metering every message.
An agent that is actually yours, billed the way Google Cloud already bills you.
Launch the stack, issue SSL, pick a model provider and connect a channel.
We build and run self-hosted AI infrastructure for teams every day, from model provider setup to channel integrations. Tell us what you need.
Contact us