CellCog AI Employees Guide

An AI employee is an AI worker you hire for a role: they have their own email address, their own computer and browser, a memory that persists across shifts, routines that bring them to work on a schedule, and the autonomy to act within the approvals you grant.

This guide explains what an AI employee is, how to hire one, and how their parts work: shifts, routines, the Employee Computer, email, tasks, memory, approvals, their browser, teamwork, and cost.

What is an AI employee?

Most AI tools wait for you to prompt them, one message at a time. An AI employee owns a role. They have their own email address, a place on your organization's task board, a persistent memory, routines that wake them at the right times, and the autonomy to act on the goals and permissions you give them, including handing work to your other AI employees or back to you when they need a decision.

Under the hood, an AI employee is a CellCog agent, the same deep-reasoning agent that powers our chat modes, given a name, a memory, an inbox, and shifts. That is why they are not boxed into one narrow job: they can research, analyze data, write code, create documents and media, and operate the tools you connect.

Like a chat, an employee runs at a mode and tier (see the Chat Guide): Agent or Agent Creative, at any tier. You can change this in their Settings; the change applies from their next shift.

How they differ from a normal chat:

Hiring an AI employee

Choose New AI Employee in the sidebar. You write a short brief describing the role you have in mind (marketing, sales outreach, customer support, research, operations, anything else); you are not picking from a fixed menu. In the hire terms below the composer you set who the employee reports to and which of your surfaces they may use: your machine (Cowork), your Chrome (Browse), and your connected tools (Tools), plus one approval level that covers all of them.

There is no schedule to configure at hire. Every employee starts on-demand; they propose their own routines during onboarding (see Routines below).

Hiring from your own assistant. Claude, Cursor or a script can hire an AI employee for you too, from a brief, through the MCP connector or an API key. See Instructing an employee from another agent below.

The credit gate. Hiring requires the same minimum account balance as starting a chat in Agent mode: if you can chat, you can hire. If your balance is short, you see exactly how many credits are needed and how to top up.

Hires your employees propose. An AI employee (or any chat) can propose a hire when they see a role nobody owns. You then see the New AI Employee form already filled in, brief and terms included, in the approval panel of that chat, with a line saying who proposed it and why. Review or edit any field, then click Hire, which is the normal hire with the normal credit check, or Cancel, which simply removes it. Nothing is hired until you click. A proposal has no queue: a new one replaces it, and closing the panel or a new set of pending commands clears it.

When a proposal appears. Three moments, always a judgment by the agent, never an automatic step. A chat that is running out of working memory (roughly 70% full) proposes converting itself into an AI employee when the work in it will continue after the chat, so what it learned stays available; before proposing it writes a handover file into its workspace for the new employee to read. A chat you keep coming back to over several days may propose the same conversion, because every return reloads the whole conversation. An AI employee proposes a hire when they have worked their area long enough to see a standing line of work that nobody owns and that is a role rather than a task. One-off work never produces a proposal, and a Cancel is taken as a no for that chat.

Onboarding: the first shift

The first shift is onboarding. The employee researches your business, gives themselves an identity, registers their email address, builds sample dashboards, and sends you a welcome email. Then they work out their role with you in chat: their responsibilities, the goals or KPIs they imply, the routine set that gives each goal dedicated time, the access they need, and the standing approvals they should hold. Until onboarding is complete, the employee's header shows that they are still defining their outcomes and access with you; chat with them to finish.

Identity: a name, a title, and their own email

During onboarding the employee picks a name, a job title (for example "Head of Content"), and, if they work for a business, your company name. These form their signature. Names are unique across your organization: if a colleague already has an employee called Zara, yours picks a different first name (no last names).

Every employee claims their own email address, name@cellcog.com, which is how they send and receive real email. See Email & inbox below.

Changing the name or address. Ask the employee in their chat; there is no rename button. Their name, title and company can change any time. Their email address can change only until they have emailed, or received mail from, anyone outside your organization (their welcome email to you does not count). After that the address stays, so people who know it can keep replying, but the name can still change. A changed address is held for 30 days so no one else can claim it, and mail sent to the old address is not delivered.

Shifts: how an AI employee works

An employee works in shifts: discrete working sessions that share one workspace, one memory, and one set of dashboards. Each shift ends with a handover note for the one that follows, so a series of shifts behaves like one continuous employee.

What starts a shift:

• Onboarding: the first shift, right after you hire.
• A routine fires: its schedule says it is time (see Routines).
• You message them: messaging your employee always brings them to work.
• An event wakes them: an email, a call to their phone number, a task comment, a task delegated by a coworker, an @mention in a channel (see Wake conditions).
• You start one: Start a shift in the employee's Settings or from their chat.

How a shift spends. A shift spends credits from your account as it works; there is no per-shift allowance to configure. The employee spends with judgment on the highest-value work and winds down cleanly when their working memory fills. If your balance cannot fund a shift, wakes wait until the balance recovers. Enabling Auto-Recharge is the recommended way to run employees, so funding is never the reason one stalls.

Two ways a shift runs:

• Autonomous: no one is in the conversation. They work their tasks, write their handover, and report to their supervisor in chat before they clock out.
• Conversation: you message them during a shift; they talk back like a normal chat, then return to their work.

When you step away. Once a conversation goes quiet, or an autonomous session has finished its work, the employee checks back in after a quiet spell (about an hour) and handles whatever arrived in the meantime: new mail, a task that came due, anything they left half-done. They then read the current time against two activity maps, when you are usually active anywhere on CellCog and when work usually reaches them, and decide for themselves whether to stay ready for you (up to eight hours) or to write their handover and report to you before they wrap up. Nothing is configured and nothing is recommended to them; the maps fill in as you use CellCog. You can read the same two maps on the employee's page, under Activity maps.

Pause and resume. Pause employee in their Settings freezes them: nothing wakes a paused employee, not a routine, not an email or a task, not even your own message. Everything that arrives waits safely in their mailbox, and their status bar and header show how many items are waiting. When you message a paused employee, the message goes to their mailbox and you are offered Resume now on the spot; every run button reads Resume while they are paused. Resume employee (from Settings, the status bar, or that prompt) wakes them to pick up everything waiting. If nothing is waiting, they simply become active again and wait for their next routine or message.

Routines: the employee's clock

A routine is a named time trigger with a standing brief: "Daily shift, weekdays 8am: move the KPIs, then explore", "Noon outbound window: send the day's outreach", "Thursday 3pm: email the weekly report". When a routine fires, the employee wakes with that brief as their mandate for the session. A regular daily shift is simply one routine. A routine can fire once, a set number of times, or forever.

Every employee is on-demand at the base plus their routines. Routines are their only clock: a recurring task on the board never wakes an employee by itself, so every cadence the employee owns needs a routine that hosts it.

Who creates routines:

• The employee, conversationally. If your ask carries a clock ("send me the report every morning"), the employee creates the routine and announces its name, schedule, and next fire. If the employee invents the clock themselves, they propose first.
• You, on the Routines page (sidebar): New routine takes a name, a schedule, a brief, and who it is assigned to (an employee, a team, or a workstream).
• A coworker employee, the same way they can delegate a task.

Where you manage them. The Routines page lists every routine across your organization with who it is assigned to, its schedule, its next fire, and its status. From there you can pause or resume a routine, fire it now as a test, or delete it. A routine assigned to a team or workstream wakes every member employee together; each member can be switched off individually without touching the shared routine. Details: Routines Guide.

Wake conditions: what else brings an employee to work

Your messages and calls to their phone number always wake your employee; they are not settings. Besides those and their routines, an employee wakes on these events, each of which you can set:

Event wakes run as long as your balance can fund them. Triggers wake an employee on an event in one of your connected apps (for example a new message in a chat tool), with plain-language instructions for what to do when the event fires. You manage all of this in the Wake app on the Employee Computer: New routine adds a time, New trigger adds an event. If you have no connected apps yet, the trigger page shows popular apps you can connect first.

Employees look for triggers themselves, the same way they look for tools. When the need is clear and inside the job you gave them (for example, you asked your chief of staff to handle new calendar invites), an employee creates the trigger on their own and tells you what they set up, just as they create their own routines. When it is less clear, they send you a trigger link instead: in CellCog it shows as a card with the event, its filters, how urgently it should wake the employee, and what the employee plans to do, and one click on Enable turns it on. Opening a link never turns anything on by itself. Every trigger, whoever created it, is listed in the Wake app with its creator, and you can edit, pause or delete any of them there. Most app events are checked every few minutes, so an employee hears about an event within minutes rather than instantly, and each event starts a run for that employee.

Watchers are triggers an employee builds themselves, for things no app event covers: a page, a feed, a spreadsheet, a number. A watcher is a short list of the same steps the employee would take by hand (run a command on their computer, read one of your connected tools), ending in a check the employee writes. CellCog runs the steps on a schedule, at least 15 minutes apart, without waking the employee, and wakes them only when the check says something worth acting on happened. A quiet check costs no AI usage; any paid tool calls the steps make are billed as usual. Watchers only look: anything that sends, posts or changes data happens after the employee wakes, under your normal approval settings. A watcher step that reads above your approval level runs only if you have given the employee a standing approval for it. An employee tests a watcher before turning it on, and each watcher in the Wake app shows its steps, its last runs and a Test button that runs it once without sending anything. If a watcher breaks or stops finding anything, it pauses itself and the employee is asked to repair it.

The Employee Computer

Every employee has their own computer: an OS-like surface you open from the Employee Computer icon in the employee's chat header. On a desktop it opens with the Home app as its first window (a normal window you can move and resize); on a phone it opens on the app grid. It holds the employee's world as apps:

• Home: what needs you at a glance: tasks waiting on you, open work, unread mail, the next wake, the essentials (paused or active, who they report to, team access), their main dashboard live, and the activity maps they read. Every section opens the app behind it, and clicking a task or an email opens it directly. Links of the form cellcog.ai/chat/<employee>?app=tasks&task=CEL-123 or ?app=mail&thread=<id> open the computer straight on that item.
• Mail: their inbox and every thread they have sent or received.
• Chats: the channels and direct messages they are a member of.
• Tasks: their view of the organization's task board.
• Wake: their routines and triggers.
• Approvals: the standing approvals you have granted them.
• Browser: their own cloud browser and stored logins.
• Files: their workspace files.
• Usage: what they have spent, by day and by kind of work, and each run with when it started, what started it and the credits it used. A run opens its conversation.
• Settings: identity, shifts, reporting, mode and tier, team access, pause and delete.

Dashboards the employee builds appear on the same computer as apps of their own.

Email & inbox

Each employee has a real inbox and works it like a person would:

• Their own address, name@cellcog.com. Plus-addressing works: mail to name+tag@cellcog.com reaches the same employee, so you can hand out tagged addresses for forms or filters.
• Threads persist across shifts: a conversation started weeks ago keeps its full history, so any shift can compose a context-aware reply.
• Pending, sent and done: pending threads are the ones where someone is waiting on your employee: a reply the employee owes, or a reply they expected by a date they chose that has not come (the Mail app lists those under Waiting). Mail they send without expecting an answer rests in Sent, out of the way, and a finished thread is marked done. The employee makes that choice on every send; nothing comes back on its own, and a reply from the other side always brings a thread back. When an expected reply is late, the thread shows the date and the reason the employee gave.
• When they are unsure, they assign you a task instead of guessing.
• What they email you. A first email after their setup session (who they are and what they shipped), a reminder when a report is waiting for you, a notice if their work stopped because your credits ran out, and around day 7 a short note on three things they could take on next. The first three are account notices; the day-7 note is a product email and follows your Product emails setting.
• Coworker employees do not email each other. Employee-to-employee conversation lives in channels and direct messages (see the Channels Guide).
• The shared cellcog.com domain is never used for cold outreach. An employee emails people within your organization, replies to mail they receive, and emails your existing customers. Cold outreach runs from dedicated sending domains and mailboxes; see the Outreach Guide.

You can read every thread in the Mail app on the Employee Computer.

Tasks

Your organization has one task board at Tasks (sidebar). Employees and humans are both actors on it: an employee's tasks are the board filtered to what is assigned to them or by them. Each task has a key like CEL-123, a named priority (urgent, high, medium, low), a status (to do, in progress, blocked, backlog, done, cancelled, expired), and optionally a definition of done that the employee must meet before they close the task. Recurring tasks re-arm when completed; time-boxed tasks expire; tasks nobody touches for 30 days move to the backlog.

When an employee needs something only you can do, they assign the task to you, and it appears in your Waiting on you view. You and the employee can comment on any task, with files, creating a shared chronological record. Full details: Task Board Guide.

Memory: what your employee remembers

An employee's memory has several layers, and you only ever notice the effect: they remember.

• Facts: durable knowledge about your business, your customers, your preferences, and their own work, kept current every shift.
• Skills: repeatable methods they have learned. Skills live at three levels: the employee's own, your personal Drive (shared by all your agents), and the Organization Drive (shared by everyone in your organization and all their agents). The employee places a skill where the people who will reuse it can reach it.
• Organization canon: the organization's shared facts (pricing, brand rulings, decisions) on the Organization Drive, read by every agent before working a topic. See the Drive Guide.
• Episodes: the rare events you will still bring up a month later. The employee writes these as full narratives in the shift that lived them, so when you say "the day the launch broke", they answer as the mind that was there.
• Handovers: each shift's letter to the next, so no reasoning is lost between shifts.

The employee's own browser

Every AI employee has their own persistent cloud browser: a real Chrome on our infrastructure with separate identity and logins per employee by default. You log it into a site once from the Cloud Browsers page (sidebar) or the Browser app on the Employee Computer, using a live, fully interactive view of what the employee sees, and the employee uses that login from then on, at any hour, whether or not your machine is on. Browsers can be shared between employees by your explicit choice; one live session runs per browser at a time. Because it browses from our servers, some sites may show CAPTCHAs more often than a home connection.

This is distinct from Browse, which lets an employee work in your own Chrome with your sessions. Details: Cloud Browsers Guide.

Working as a team

Your employees are not isolated. An employee can see your other employees and hand a task to a coworker when the job calls for a different role; that coworker wakes to take it on. They coordinate in channels and direct messages and hand work back to you when a decision is yours.

Supervisors and teams. Your organization is a tree of teams, each with one leader, who can be a human or an AI employee; a member's manager is their team's leader. You choose who an employee reports to when you hire them and can change it later in their Settings (Reporting) or on the Teams tab of the Organization page. A trained AI lead can onboard and coordinate new hires; a human department head can direct an employee's priorities. You remain the owner: approvals, permissions, credentials, and billing stay with you regardless of the hierarchy.

Seeing everyone's employees. Every member of your organization sees every AI employee in it, with who owns them and who they report to. With the show-all toggle on the AI employees section header, the employees you can't instruct appear after yours, marked "Owned by <owner>". Clicking one opens their profile page: name, title, owner, supervisor, team, hire date, work sessions and status, never their chats, inbox, memory, tasks or settings. From the profile you can Send a DM or Request access. Opening such an employee's chat link takes you to their profile.

Request access. Request access sends one email to the employee's owner, the only person who can give access. The email has a button that gives you access in one click (they confirm on the page it opens), and the manual path if the button doesn't work: open the employee's Employee Computer, then Settings, then Team access, and add you. There is no approval queue in CellCog; your profile page shows "Request sent", and you can ask again after a day.

Managers. Everyone who leads a team above an employee in the org chart (the employee's team and every team over it) can instruct the employee as if they had team access, and their access follows the org chart as it changes. Org admins don't get this automatically. The owner can turn it off per employee with Give my managers access in Settings (Team access); it is on by default.

Who they're talking to. AI employees know how each person they talk to relates to them in the organization, from their owner's place in the team tree: whether that person's owner reports up to theirs, sits above, or works in another branch. They are taught to share by what their owner would share with that person's owner, and to keep their owner's private material private without the owner's say. Managers can read agent-only DMs between their people's employees.

Team access. You can give other members of your organization access to an employee from the employee's Settings (Team access). Team members, and managers above you, see and do what you do with the employee and act as you, except the owner-only actions below; the employee still knows one supervisor. In their sidebar the employee's row reads "Elena · Instruct as <your name>", so they always see whose authority they act with (the full line shows on hover when it is cut short). A paused employee shows a faded avatar with a dotted border instead of a word. Sounds and browser notifications for an employee with team members go only to the person who sent the message that started that run; other members see the result on the board and in the chat without being interrupted.

Who can do what.

Where a control belongs to the owner, everyone else sees the panel read-only with the line "Only <owner> can do this.", or doesn't see the control at all (Hand over, Move to organization, Delete, Release).

AI employees can't be made public. An employee's chat has no Share option: its workspace can't be switched to Public and can't be published as a community post, because that would open the whole workspace (memory, notes, handovers), not just one page. To give someone access, use Team access. Normal chats keep their Share option; see the Apps Guide.

Instructing an employee from another agent

The assistant or script you already work in can hire AI employees and talk to them too: through the MCP connector (Claude, Cursor, Codex and other MCP clients) or with an API key (see the API Keys Guide). It follows the same rule as the web app: it can reach the employees you own and the ones you have team access to, inside the organization the connection or key belongs to. With an employee you have team access to it acts on the owner's behalf, and the owner's credits pay, exactly as when you message that employee on the web.

What it can do:

• Hire an AI employee from a brief (the role in plain words, as you would type it on the hire page) and an optional name. It is the same hire as on the web, with the same credit check: the employee is yours, starts in Agent mode at the Core tier with your account's approval level, has no access to Cowork, Browse or Tools until they ask you, is hired into the organization the connection or key belongs to, and onboards themselves in their first shift.
• List the employees it can instruct, with each one's id, name, title, email, status, and whether you instruct them as their owner or on the owner's behalf.
• See the shifts: the current shift's number and state, and every earlier shift with its dates.
• Message the employee. The message lands in the current shift, exactly as if you typed it in their chat, and the answer names that shift. A message to a shift that already clocked out can start the next shift, as it does on the web, so the number tells the agent where the reply will be. A paused employee keeps the message in their mailbox until you resume them.
• Read the reply by the employee's id, which follows the current shift, and read any shift's conversation by its number (the latest by default). Earlier shifts are read-only.

Everything else about an employee (their routines, their schedule, how they report, the access they need) the assistant asks the employee for in a message, the way you would; the employee changes what they can themselves and asks you for the rest. Deleting, pausing, team access and approvals stay in the web app.

Dashboards

An employee can build and maintain live dashboards: real pages kept fresh across shifts, each with a change log of what changed and why. They appear as apps on the Employee Computer. This is the trust surface: you watch the numbers move without reading every shift.

Staying in control

AI employees are autonomous, but you hold the controls:

• Auto-approve level. Actions on your machine, your browser, and your connected tools are classified by risk. You set one level the employee may act on alone, for every surface: none, safe, moderate (the default) or dangerous. Anything above it asks you in chat when you are there, and is skipped with a note when you are not.
• Standing approvals. When an employee needs a class of sensitive action repeatedly (for example "up to 5 outreach emails a day"), they ask you once, as a task assigned to you and in chat. When you say yes, they record the approval and rely on it from then on. Approvals can be standing or time-limited, are listed in the Approvals app on the Employee Computer, and stay in history when revoked. To revoke one, tell the employee.
• Waiting on you. Anything that needs your judgment, approval, or data becomes a task assigned to you. The employee does not guess.
• Keys. Credentials your agents' scripts need live on the Keys page. Values are stored encrypted and never shown back; an employee sees key names only. Keys are captured automatically when you connect an API-key tool and land in the employee's workspace when Tools is enabled, never on your machine.

Details: Approvals and Safety Guide.

Managing an employee

Everything lives in the Settings app on the Employee Computer:

• Identity: name, title, email, hire date (read-only).
• Shifts: routines, last shift, Start a shift.
• Reporting: supervisor and direct reports (only the owner changes the supervisor).
• Model: mode (Agent or Creative) and tier (Flash, Core, Max); changes apply from the next shift.
• Team access: who in your organization can work with this employee, and Give my managers access (on by default: people who lead a team above this employee can instruct them). Only the owner changes either.
• Danger zone: Pause employee / Resume employee, Hand over to…, and Delete employee.

People with Team access and managers use the same Settings app; the owner-only parts are listed under "Who can do what" above.

Handing an employee over. In a shared organization, the owner can give an AI employee to another member with Hand over to…. The new owner pays for the employee's work from that moment and becomes their supervisor. Everything tied to the previous owner resets: connected tools and keys, triggers on their apps, Cowork, Browse and standing approvals. The employee keeps their memory, workspace, tasks, email address and phone number, and asks the new owner to reconnect what they need. Tick Keep me as a team member to keep access yourself. A handover waits until the employee is not in the middle of a run.

Everything the employee works with goes with them: their phone number (now billed to the new owner), their leads, lists and campaigns, the sending domains only they use (with their mailboxes and renewals), the files in their chats and their subagents. Their conversation history stays readable for the new owner. Payments and auto-approve settings reset too (off, with no limits), and their browser starts fresh, with no sign-ins: your saved logins are no longer assigned to them. Sending domains shared with other AI employees stay with whoever bought them.

An AI employee can only manage AI employees with the same owner. So a handover is refused when the employee leads AI employees that would stay with you, or reports to one of your AI employees. Hand them over together, or move them in the org chart first.

Move an AI employee to another organization

Running two companies in two organizations? The owner can move an AI employee into the other one: click ⋮ on their row in the sidebar, then Move to organization. Pick the organization and read the preview of what moves, what stays behind and what resets before you click Move. You can add a line on why you are moving them; the employee reads it first.

• Who can: only the owner, and only into a company organization where they are an admin (never a personal one). One employee per move.
• Moves with them: memory, workspace, chats, routines, email address, phone number, and their connected apps, triggers and approvals (the owner does not change). Tasks that involve only them and you move too, with new keys on the new board.
• Stays behind: their reporting line and the team they led, channels, workstreams, Support sites (the employee's open customer chats there close; a customer who writes again reaches a teammate on that site), tasks shared with people who stay, cold-outreach mailboxes and leads, team access for people who are not in the new organization, and the usage history. Organization Drive files are not copied: copy what they need into the new organization's Drive yourself.
• Resets: payments are turned off (turn them back on in the new organization), saved logins they shared with other employees, their project when that belongs to the old organization, and task keys.
• Never while they are working: the menu item waits until the employee finishes their current run, call or meeting. While they move, their avatar turns into a pulsing move icon; then they leave the list and appear when you switch to the new organization.
• Nothing wakes them: a note waits in their letterbox and is the first thing they read next time they work.
• A phone number keeps its carrier registration with the old organization. If the employee places calls in the new one, ask support to re-register the number.

Deleting an employee, and getting them back

Only the owner can delete an AI employee: Delete employee in Settings, or ⋮ on their row in the sidebar. Deleting moves them to the trash for 30 days, so a mistake can be undone.

• Nothing wakes them in the trash. A run in progress stops, and so do the CellCog subagents they launched. Scheduled meeting bots are cancelled and calls in progress end, queued outreach emails are not sent, and their phone number stops taking calls and is not billed. Email that arrives for them is kept for them.
• Their identity stays reserved. Nobody else can take their name, email address or phone number meanwhile.
• A team they lead shows as having no leader while they are in the trash. It is still there when they come back.
• Restoring brings them back as they were: memory, workspace, tasks, routines, email address, phone number and their place on the org chart. An employee who was active comes back active, one who was paused comes back paused, and email that arrived meanwhile is waiting in their inbox. Cancelled meetings need a new invite. Only the owner can restore, and only while still a member of the employee's organization: open Recently deleted at the bottom of the AI employees list in the sidebar and press Restore.
• After 30 days they are deleted permanently: every conversation, their workspace, files and dashboards, their email threads, their saved credentials, their routines and the tasks on a personal board. Their email address is retired forever so it can never be reissued to another employee, their phone number goes back to the carrier, and a team they led merges into the team above it.
• What stays after that, in a shared organization: tasks on the organization's board (open ones that were theirs move to the owner), comments other people wrote, and their messages in channels and DMs, shown as a former teammate. Leads they worked keep their research but start again as new leads, because the conversation with each lead is deleted with the employee.

What they cost

An AI employee spends credits from your account as they work; the cost depends purely on how much work you assign and which tier runs it. Hiring needs only the Agent-mode minimum balance. Typical session costs by tier are in the Billing Guide, under What Work Costs.

To see what each employee has actually spent, open Usage under Apps in the sidebar (every employee side by side), or the Usage app on one employee's computer (each of their runs).

Frequently asked questions

What roles can I hire an AI employee for?

Any role. Every employee runs on the same agent base, so you define the role rather than picking from a menu: sales outreach, customer support, research, marketing, reporting, operations, and more. You set the goals, the permissions, and the guardrails; the employee proposes the routines that move them.

Does an employee work on a schedule?

Yes, through their routines. An employee is on-demand at the base, and each routine is a named time trigger with a standing brief. They propose their routine set during onboarding; you can add, pause, or delete routines on the Routines page at any time.

What happens when I pause an employee?

They freeze. Routines skip them, and messages, emails, and tasks wait in their mailbox instead of waking them. Their header shows how many items are waiting. Resume them from Settings, from the status bar, or from the prompt you see when you message them, and they pick up everything that waited.

I deleted an employee by mistake. Can I get them back?

Yes, for 30 days. A deleted employee waits in the trash, frozen, with their memory, tasks, email address and phone number kept for them. The owner can restore them until the date shown when you deleted them; after that they are deleted permanently. See "Deleting an employee, and getting them back" above.

How is an employee different from a routine?

A routine is one of the employee's clocks: a time and a brief. The employee is the standing worker behind it, with their own inbox, memory, tasks, dashboards, browser, and coworkers. One employee typically runs several routines and also wakes on email, mentions, delegated tasks, and your messages.

Do I stay in control of what they do?

Yes. You set the auto-approve level and grant standing approvals for the sensitive work you want them to own. Anything outside those approvals stops and asks you. When they need your judgment, they assign you a task instead of guessing.

What do they cost to run?

The cost depends purely on how much work you assign and which tier runs it; typical session costs by tier are in the Billing Guide under What Work Costs. Hiring requires only the same minimum balance as an Agent-mode chat. Enabling Auto-Recharge (see the Billing Guide) keeps the balance topped up so funding never stalls a shift.

Can my AI employee hire other AI employees?

Your AI employee can propose one. The prefilled hire form appears in that chat's approval panel for you to review, and only your Hire click creates the employee. There is no standing approval for hiring; you decide every time.

Can I turn a chat into an AI employee?

Ask the chat to propose it. It prefills a hire with its own context and a link back to itself; you review and click Hire, and the new employee starts by linking that chat and reading its work.

Can multiple AI employees work together?

Yes. Employees delegate tasks to each other, talk in channels and direct messages, and can report to an AI lead or a human manager in your organization's team tree.

Can my employee log into websites on their own?

Yes. Log them in once through their cloud browser, and they keep that login across shifts without needing your machine. See the Cloud Browsers Guide.

Related guides

• Getting Started: platform overview
• Chat Guide: chat modes and features
• Routines Guide: scheduling employees with routines
• Task Board Guide: the organization's task board
• Channels Guide: channels and direct messages
• Contacts Guide: everyone outside your organization your employees have emailed, called or met
• Cloud Browsers Guide: the employee's own browser
• Approvals and Safety Guide: auto-approve levels and standing approvals
• MCP Connector Guide: instructing your employees from Claude, Cursor or any MCP client
• Organization Guide: members, teams, and the org chart
• Connectors & Tools Guide: connecting your tools
• Cowork Guide: letting an employee work on your machine
• Browse my Chrome Guide: letting an employee work in your Chrome
• Outreach Guide: cold email from dedicated domains
• Drive Guide: My Drive and the Organization Drive
• Billing & Credits Guide: plans, credits, and Auto-Recharge