Viktor.comAn AI coworker that actually does the work
Promoted
myHermy is a personal AI agent that lives on a private cloud server you own — always on, reaching out on WhatsApp, Telegram, Discord & Slack. Connect the ChatGPT, Claude, Copilot, or Grok plan you already pay for. Zero setup, daily backups, your data stays yours.
Interactive
Payment Required
Launch Team / Built With
Mailwarm
How do you handle message history privacy across those channels, what gets stored on the server vs only forwarded to the model?
@naimz Hey Naim! Great question! privacy was a core design decision, so happy to walk through it.
The short version: nothing sensitive lives on our servers.
Every agent runs on its own dedicated VPS that you own — fully single-tenant. All the channel bridges (Telegram, WhatsApp, Discord, Slack, Signal, email) run on your box, so messages flow channel → your agent → model. They never route through our control plane.
What's stored where:
Conversation history lives locally on your VPS, never in our database.
Our database only holds account + infra metadata: server config, billing, and encrypted-at-rest credentials (bot tokens, OAuth). Zero message content.
Model calls go directly from your VPS to the provider (Anthropic/OpenAI/etc.), we never see the prompts or responses.
Why history is stored at all: it's what makes the agent actually useful — the memory the model reasons over for multi-turn context, follow-ups, and long-term personalization across sessions and channels. The important part is where it sits: on your own server, not ours. So you get a stateful agent and full ownership - inspect, export, or wipe it anytime.
One honest caveat: daily backups are full-disk snapshots, so they do include that local history (7-day rolling retention, same datacenter, deletable anytime from your dashboard).
So "forwarded to the model" = everything; "stored on our servers" = essentially nothing beyond your account.
Let me know if you want to give it a try, will be happy to provide you a promocode for 40% discount on first month.
@naimz Hey Naim! Great question! privacy was a core design decision, so happy to walk through it.
The short version: nothing sensitive lives on our servers.
Every agent runs on its own dedicated VPS that you own i.e fully single-tenant. All the channel bridges (Telegram, WhatsApp, Discord, Slack, Signal, email) run on your box, so messages flow channel → your agent → model. They never route through our control plane.
What's stored where:
Conversation history lives locally on your VPS, never in our database.
Our database only holds account + infra metadata: server config, billing, and encrypted-at-rest credentials (bot tokens, OAuth). Zero message content.
Model calls go directly from your VPS to the provider (Anthropic/OpenAI/etc.) — we never see the prompts or responses.
Why history is stored at all: it's what makes the agent actually useful, the memory the model reasons over for multi-turn context, follow-ups, and long-term personalization across sessions and channels. The important part is where it sits: on your own server, not ours. So you get a stateful agent and full ownership - inspect, export, or wipe it anytime.
One honest caveat: daily backups are full-disk snapshots, so they do include that local history (7-day rolling retention, same datacenter, deletable anytime from your dashboard).
So "forwarded to the model" = everything; "stored on our servers" = essentially nothing beyond your account.