Your installation

Player One runs as your own installation: one copy of the platform at your own web address, holding your chats, your agents, your files, your memory and the accounts you have connected. Nobody else works in it and nothing in it is mixed with another customer's. Change a setting here and it changes for you alone.

Your installation is kept up to date for you, and you are told what changed rather than left to notice. When a new version arrives a red Required update banner appears and your assistant opens a conversation that goes through each release in turn, explaining what is new and repairing anything the new version needs. Reloading the page returns you to that same conversation. If the moment is wrong, Skip update puts it off and the walkthrough comes back later.

Which version you are on is shown on Admin > Runtime, and it is the first thing to quote if you ever report a problem.

Signing in

Your installation has its own sign-in page at its own address, separate from your account at 1p.dev. Once you are in, the browser stays signed in while you use it:

• Open as many tabs as you like. They share one sign-in, and one tab's trouble renewing it does not sign the others out.
• A tab you are using keeps the sign-in fresh, so a long working day does not end at the sign-in page. A browser nobody has used for a day asks you to sign in again.
• A dropped connection or a slow moment on the server does not sign you out. Only a sign-in your installation has refused does.
• Updates and restarts of your installation do not sign you out either.

Each browser you sign in from is a device on Admin > Devices (/admin/devices). Sign out this browser signs out the one you are using, which is the thing to do on a shared computer. Forget removes a device you no longer use, which can sign in again later; Revoke bans one for good, for a computer that is lost or not yours; and Forget stale devices clears everything not seen for 14 days. When a computer you already had is set up again from scratch, its request says it is the same machine, with Approve and replace for the old pairing or Approve as a new device.

The Admin area

Everything you configure or supervise about your installation lives in Admin, which you open from the bottom of the sidebar.

It replaces the sidebar with a darker menu of its own, and Back to Player One at the top takes you back to your chats. Overview (/admin) comes first, with your installation's name, whether it is connected and the version it runs, and a link to every Admin page. Below it the pages are grouped as Installation, Models, Communications, Security, Memory and Developer, with one group open at a time. Every page has its own address, and so does every tab within a page, so a link lands on exactly the view it names.

Two switches on Admin > Runtime (/admin/gateway) change what the menu shows. Advanced adds Chat Index, System Logs and the Voice Harness. Development adds two diagnostic pages and is off on the hosted product. Both are remembered in your browser and change nothing for anyone else.

The pages worth knowing

Page	Where it is	What you do there
Branding	Installation, /admin/branding	Your installation's name and description, its logo, app icon and three colours: what everyone who opens it sees.
User	Installation, /admin/user	Your own name, role and picture, and the light or dark theme, which follows this browser only.
Runtime	Installation, /admin/gateway	Connection status, component versions, the two switches above, and the buttons to clear caches or restart.
Usage	Installation, /admin/usage	Where the spend is going, by model and by agent.
Resources	Installation, /admin/resources	Live health of the machine your installation runs on.
Backups	Installation, /admin/backups	Export a project or chosen items, and import an archive from elsewhere.
Providers	Models, /admin/providers	Connect your AI accounts and set the order they are tried in: see Models and providers.
Agent Models	Models, /admin/models	Which model answers, what it falls back to, how hard it thinks, and rules for particular work: see Models and providers.
Image Models	Models, /admin/images	The model that draws pictures and the one that describes images a chat model cannot see, kept separate from the chat models.
Embedding Models	Models, /admin/embedding-models	The model that turns text into embeddings, so text can be compared by meaning.
Voice Models	Models, /admin/voice	The models that speak, narrate and transcribe, and whether voice is healthy: see Voice.
Judgement Models	Models, /admin/judgement-models	The model your agents ask for quick typed judgements, once a judgement provider is connected.
LLM Endpoints and API keys	Models, /admin/endpoints; Security, /admin/api-keys	The address your own apps call to reach your models, and the keys they and your coding agents use: see Use Player One from your own apps.
Agent Sharing	Communications, /admin/agent-sharing	Share an agent with another installation, use one shared with you, or open a sandboxed agent to the public: see Share an agent.
MCP	Communications, /admin/mcp	Connect a coding agent such as Claude Code, Codex or Cursor to your installation: see Build with it.
Messaging	Communications, /admin/channels	Connect, test and disconnect messaging and email: see Channels and communication.
Phone lines	Communications, /admin/phone-lines	Your agents' telephone lines and the calls they have taken: see Voice.
Webhooks	Communications, /admin/webhooks	Addresses an outside system can call to start a flow or an agent.
Alerts	Communications, /admin/alerts	What the platform tells you about, where it reaches you, and what it has already sent.
Vault	Security, /admin/vault	Passwords and keys your agents can use without ever seeing them: see Vault and sandboxing.
Devices	Security, /admin/devices	Every browser and computer signed in to your installation. Approve a new one, rename, forget or revoke one, or sign this browser out.
Skills	Memory, /admin/skills	The catalogue of what your agents know how to do: browse, search and preview: see Skills.
Fold	Memory, /admin/fold	Indexing across everything searchable: health, contents, and pause, resume or reindex: see Memory.
Engram	Memory, /admin/engram	The tidying your memory gets while you sleep: what is queued, what ran, and when it runs next.
Chat Index	Memory, /admin/search	The cache behind your chat list and the index your agents search, with the rebuild buttons. Behind the Advanced switch.

Usage

Usage answers the question of what this is costing you.

Its first tab totals the tokens and the money for a window you pick, today or the last 7, 30 or 90 days, charts them over time, and breaks them down by model and by agent, so you can see which agent and which model are doing the spending. The second tab, API calls, lists the individual calls your own apps made through your endpoints, one row each with the model, the size, the cost and the outcome. Tokens, cost and caching explains what the numbers mean.

Backups

Backups is how work leaves your installation and arrives in another.

Choose a project and export it as an archive: its agents, workspaces, Memory Spaces, flows, jobs, scheduled tasks, contact lists and chats, with workspace files, private agent skills, chat history and secrets as separate boxes you have to tick. Importing works the other way: upload every part of the archive, pick what you want from it, and review the plan before it is applied. Existing items are kept unless you choose to overwrite them. Chats travel too if you tick them, which is off by default because they are usually the largest part. Connected accounts never travel, so a restored channel or scheduled task waits for you to reconnect it. Every export and import is recorded. Back up and move your data has the steps.

Resources

Resources is the health of the machine your installation runs on, and it only watches: nothing on it stops or changes anything.

You get memory against its limit, processor use, a bar for each disk split into what is filling it, and history you can wind back thirty days, with markers where something restarted or a browser was released. The one button, Clean up storage, asks your assistant to do the work, and it tells you what it plans to delete before deleting it. The same reading drives a small gauge in the top bar of every page, and a banner across the site if things get tight.