Built on Hermes — the open-source agent that grows with you
An AI agent that actually remembers you.
Every AI chat starts from zero. FoundryStack gives you a private agent that doesn't — it keeps its memory, builds its own skills from experience, and is ready the second you sign in. No servers to run. No Docker, DNS, or API keys to wire up yourself.
Your private agent is provisioned automatically after you sign in — usually ready in a couple of minutes.
You shouldn't have to re-explain yourself to AI every single time.
Stateless chat forgets everything
Close the tab, lose the context. Every new conversation starts with a blank slate, no matter how much you've already taught it.
Self-hosting is a real project
Running your own agent means a VPS, Docker, DNS, TLS, and an identity provider that actually works with your login — a multi-day build even for engineers.
Shared tools mean shared risk
A generic assistant with no isolation boundary means your data, memory, and usage sit alongside everyone else's.
What you actually get
Persistent memory, by design
Your agent remembers past conversations, decisions, and context — across sessions, across days, without you repeating yourself.
Skills it builds on its own
Hermes creates and refines its own skills from experience. The more you use it, the more capable it gets — automatically.
Your own isolated agent
Every account runs in its own container with its own storage, its own resource limits, and its own credentials. Nothing shared, nothing crossed.
Wherever you already are
Talk to your agent from a clean web dashboard or from Telegram — same memory, same skills, same agent either way.
Sign in with Google
No new password to manage. Secure, standards-based authentication, the same way you already sign into everything else.
Zero setup, on your side
We handle the hosting, the security, the uptime. You just start talking to your agent.
How it works
-
1
Sign in with Google
We use your first name to personalize your agent from the very first message — no forms, no separate account to create.
-
2
Your agent is provisioned, just for you
A private, isolated agent spins up under your account — its own memory, its own resource limits, ready in moments.
-
3
Start working with it
Land straight on your agent's dashboard. Ask it something, give it a task, or just start the conversation you'd normally have to start over.
Get your agent
Sign in with Google. We spin up an isolated Hermes agent under your account and send you straight to the dashboard.
Capacity-capped free tier. Bring your own model key anytime from settings.
Frequently asked questions
What is FoundryStack, exactly?
FoundryStack is a hosting platform for Hermes, an open-source AI agent with persistent memory and self-improving skills. We handle the infrastructure — servers, security, authentication, uptime — so you get a private agent without running anything yourself.
How is this different from ChatGPT or Claude?
General-purpose chat assistants typically don't carry memory between conversations by default, and you're sharing infrastructure with everyone else. Your FoundryStack agent runs in its own isolated environment, keeps its own memory over time, and builds its own skills specific to how you use it.
Is my data isolated from other users?
Yes. Every account gets its own container with its own storage, its own resource limits, and no shared credentials with any other account.
What can I talk to my agent from?
A web dashboard out of the box, with Telegram support as well — same agent, same memory, regardless of where you're talking to it from.
Do I need my own API key?
No — the free tier includes a hosted model with a monthly usage ceiling. You can bring your own MiniMax (or compatible) API key anytime to lift that shared-key limit.
Is signup open?
Yes, within a global capacity cap. If we're full, you'll be queued and your agent starts automatically when a slot frees up.