Dedicated server. Your own memory. Consented context. Approval before anything is sent. Easy to set up — no terminal, no servers to rent. Safe by default — nothing irreversible without your tap. Disconnect and delete in one action, any time.
Short version. Most AI products give you a shared chat tab. Summon Agent gives you an agent that runs on your own dedicated server, with your own memory, in the messenger you already keep open. You decide what it knows. You decide what it can do. You can take any of that back in one tap.
Each Summon Agent account is provisioned on its own dedicated server in the region you pick — EU, US, or APAC. There is no shared database that mixes your prompts with another customer's. No shared model account. No multi-tenant memory pool.
No terminal. No tokens. No bots to configure. You name your agent, pick a persona and a messenger, and confirm a region. We provision the rest in the background while you fill in five fields.
Anything that touches the outside world — sending an email, paying an invoice, accepting a meeting — stages itself first. The agent shows you the draft in the messenger and waits for one tap to fire. You can require approval on absolutely everything, or relax that scope per tool over time.
Your agent only knows what you give it. The first session is a short consented setup — what you want it to know about, which messengers and tools to connect, which routines you want. Every memory item is visible, editable and exportable.
It means three things: dedicated infrastructure (your own server, not a shared pool), single-tenant memory (no shared database), and full control over context and permissions (you decide what your agent knows, what it can do, and which tools it can act on). You can export everything and wipe everything at any time.
Yours. Each Summon Agent account runs on its own dedicated server in the region you pick. There is no shared database that mixes you with another customer's prompts. We hold zero-retention agreements with model providers, and you can bring your own key if you would rather route through your own account.
Five fields. Name your agent, pick a persona, pick a messenger, pick a region, choose a subdomain. We provision the dedicated server, link the messenger, and configure secure connections in the background. You do not open a terminal, you do not paste a token, and you do not rent infrastructure yourself.
Any action that touches the outside world — sending a message, paying an invoice, accepting a meeting — stages itself first. The agent shows you the draft inside your messenger and waits for one tap to fire. Nothing irreversible runs without your tap. You can require approval on absolutely everything, or relax that scope per tool over time.
Only what you give it. The first session is a short consented setup — you choose what your agent learns about, which tools to connect, which routines to run. You can prune any single memory item, export everything as JSON, or wipe the agent and start over. Disconnect and delete is one tap.
Yes, any tool, any memory item, any time. One tap revokes a permission; the agent stops being able to use that tool in the same heartbeat. A global pause freezes every routine until you unpause. A wipe removes context, tools, history and memory in one action.
Not by default. Your server is isolated and staff do not have standing access. If you ask for support you can grant time-boxed, read-only access that auto-expires in 24 hours. Every staff session is logged and visible to you.
In priority beta, yes. If your ChatGPT Plus, Pro or Business plan is eligible, Summon Agent can use it as the AI layer while adding messenger access, memory, routines and approval-gated actions. Where it is not eligible, our managed setup or your own provider key works the same way.
Tell us who you are and what you would ask your agent in the first week.
Request a private beta invite