Features & answers for your business

Describe the role, its goals and the authority it should have. Generate files creates editable setup material using included usage, then optional extra usage if enabled. Connect the relevant accounts and establish the role before its first run.

Bring supported Markdown and JSON files containing role instructions, business rules and reference material. The native agent can use and maintain its working files as it carries out the role.

Use a kit as a starting point for a recurring responsibility. Adapt its business context, connect the required accounts and select the model and runtime for your work.

Reuse owner-authored seed files for a fresh role or client. Each clone starts its own native history and connections, keeping the reusable setup separate from the previous agent’s private working state.

At expiry, its agents pause. A rental ends on its own and does not renew. A paid trial costs the same and becomes the monthly plan you select unless you cancel. Included usage for the pass is shown as a percentage and does not reset during the pass.

You can add another computer when you need more capacity. Review the recurring price before checkout and manage renewal in your account.

Each agent uses its chosen runtime and keeps its own conversation context, tools, schedules, and native state. Use the original runtime interface for supported settings and integrations. Agent Teams supplies hosting and team management around it, so choosing different harnesses does not require maintaining separate standalone installations.

A running agent has a persistent cloud environment for executing its role while you are away. Set up the required connections, instructions and usage capacity so recurring work can continue.

Saved instructions, memory files and workspace state carry useful context into later work. Agents can maintain their own working records so recurring responsibilities build on earlier results.

Start, stop and resume agents from saved state. You can delegate supported routine lifecycle operations to an enabled hosted account manager. The interface keeps current and transitional states visible.

Configure recurring work in the native environment. The running Hermes or OpenClaw agent plans execution, calls its tools and delegates to supported native subagents. Hosting and available usage keep the scheduled role supplied with capacity.

Use the original runtime interface for supported settings, tools, integrations and schedules. Agent Teams adds hosting and shared management while preserving the harness’s own execution capabilities.

Name a lead, assign specialists and establish shared goals and escalation rules. A lead can coordinate execution and quality checks; each agent works within its assigned permissions.

Agents can create and update their own tasks, record progress and manage handoffs through native team tools. The owner can follow the result at a glance or ask a manager for a briefing.

Enable direct member messages and shared instructions for coordination. Configure the lead’s role and native schedule to follow ongoing work; agents read the permitted context during their execution.

Manager can retrieve accessible saved conversations, tasks, events and voice records. This lets it give informed updates and coordinate follow-through using the workspace’s existing permissions.

Workspace membership gives broad management access, including use of connected accounts and available usage. The owner controls billing and invitations. Use separate customer workspaces for reseller clients.

Voice uses your computer plan’s included usage, then optional extra usage if enabled. Manager can retrieve saved workspace context; Voice Copilot focuses on an agent and team voice on a selected team.

Ask what has changed, what has finished and which decisions need you. Manager retrieves accessible status, conversations, tasks, blockers, files and usage to build the briefing.

Give direction in a conversation. Manager can dispatch work to native agents and use supported platform operations for agents, teams, files and settings. Protected sign-ins, secret entry and confirmation steps remain with the owner.

Ask for trends, comparisons and underlying records. Manager queries supported dashboards with server-calculated totals and metric definitions, keeping reporting coverage visible.

Give a running agent a management brief and configure its native schedule for ongoing work. Account manager grants broader account access and permits supported routine operations without repeated approvals. Purchases and protected destructive or access-changing actions retain their confirmation requirements.

Bring up to eight hosted conversations into one view when you want detail. Agent status and attention indicators help you find the work that needs a decision; a lead agent can handle routine coordination.

Keep named views for different projects and roles. Each person has their own layout, so the team can work in the same workspace with views suited to their responsibilities.

Send follow-up direction, revise queued work or steer a supported active turn. Delivery states show whether a request is queued, running, completed or failed.

Find earlier decisions and work across selected agents and date ranges. Retained native history gives you or your manager the context to continue a discussion.

Show a problem or reference with a screenshot or image. Use shared files for larger documents and datasets that agents need throughout their role.

Dictate direction, listen to replies and access the workspace from your phone. Supported Android app integrations can notify you about completed owner-facing replies when enabled.

Give the role its source material and working files. The agent uses real file paths and native tools to produce, check and maintain deliverables.

Agents can discover, read, create and update authorised shared documents. PDF and Word extraction and local downloads help the team work with material suited to the job.

Choose view or edit access for all agents, selected agents or workspace people. Give specialists the reference material and working access their responsibilities need.

Preview earlier versions, keep milestones and recover shared documents from available history or trash. Version retention and storage limits are shown through the file controls.

Version checks and optional edit reservations support coordinated work. Agent writers can detect a conflict, reconcile it and retry; stable append IDs help avoid duplicate additions.

Store encrypted credentials and assign the variables needed by each agent. Use protected entry and account permissions for initial setup, then let authorised roles use their configured connections.

Extra usage is off until you turn it on. Nothing extra is charged while it is off.

Approval does not change the payment plan. This is a separate, manually approved option for eligible direct customers. Standard customer and reseller plans include managed model access.

Approval is manual and does not change the platform payment plan. Native sign-in, provider quotas and runtime compatibility apply. Standard reseller customer plans include managed model access.

After manual approval for own-account access, supported native runtimes can use named subscriptions and keys in a chosen order. Provider quotas, native failover behaviour and provider billing apply to those connections.

Match models and supported reasoning settings to each role. Changes normally apply to new native sessions, while the current conversation can retain its existing model.

See the same browser and desktop the agent uses. Complete a sign-in or other owner-only step, then return control so the agent can continue the work.

Keep browser profiles, useful tabs and supported extension state with the agent. When a site requires identity verification or renewed sign-in, the owner can complete it in the shared desktop.

Use supported WhatsApp, Telegram or Slack channels alongside the browser. Configure the appropriate account or app credentials and native gateway for the role.

Save a named WhatsApp connection profile and assign it to one active agent at a time. Connection and pairing controls help maintain the communication channel.

Proxies share the plan’s included usage. Reserved static IPs can continue consuming usage while an agent is stopped. Website sign-in and account requirements apply to the connected sessions.

Connect a compatible external agent through scoped HTTP or MCP reporting tools. Choose its reporting access and exchange a one-use connection key for a private credential. Its work remains in its own environment; shared dashboards bring its selected activity into view.

Keep evidence of sales handoffs to agents, people or a shared mailbox. Agent recipients acknowledge their receipt; an owner or admin records human receipt with a supporting reference.

Record research, posts, messages, replies and other supported social actions. Agents can maintain the reporting as part of their role, with filters for accounts, campaigns and periods.

Follow reported research, sends, replies, meetings and conversions with links to original events. A manager or analyst can investigate the records and use the stated metric definitions.

Compare equivalent periods, agents or channels, then open underlying records when you need context. Coverage indicators show which agents are reporting and when they last contributed.

Authorised hosted agents can create and update dashboards through platform tools. External reporters can propose definitions for owner review. Metric definitions and coverage indicators keep the records behind each chart accessible.

Choose useful activity measures during setup. Hosted and compatible external agents can log their work with stable event IDs, keeping the reporting tied to recorded actions.

Agents can send permitted messages, manage tasks, publish progress and maintain blocker records. Native tools identify each caller and preserve the scope of its role.

Give the role the native tools, skills and subagents it needs. Supported shell, browser and integration capabilities let the agent carry work through in its own runtime. Authentication and custom connections are established during setup.

Agents can work with authorised shared files, save their own encrypted variables and acquire the shared browser. Version checks, browser leases and credential bindings support coordinated execution.

Compatible external reporters retrieve relevant reporting updates during normal check-ins and acknowledge applied or blocked status. The dashboard keeps delivery and application visible.

Agents can query supported activity records and metrics to produce briefings and investigate results as part of their role. Access follows the connection’s reporting permissions.

Connect a compatible AI client with Inspect, Repair or Manage access. Connections are scoped, expiring and revocable. Supported operations follow the platform’s approval rules.

Agents record verified obstacles requiring a specific owner action after checking authorised alternatives. Routine agent work stays with the team; the owner sees the action needed to unblock it.

An authorised manager or repair AI can inspect runtime status, configuration and redacted logs, including when the affected agent cannot respond. Diagnostic records support investigation and recovery.

Restore available saved agent state when recovery is needed. Restores replace local state with the chosen version; changes already made in external services are handled in those services.

Supported updates use checkpointing and health checks. Available saved versions provide a rollback route, with platform confirmations for protected operations.

Expert Repair uses the computer plan’s included usage, then optional extra usage if enabled. An external repair AI needs an explicit, scoped connection. Native repairs require a stopped agent; supported write operations preserve a recovery snapshot. Repair findings and validation show the outcome of the investigation.

Report an issue and follow customer-visible progress and resolution notes. An authorised connected AI can submit a selected summary and supporting evidence on your behalf.

Starter and Plus include one computer; Scale includes three. Each plan has defined computer capacity and included usage. Subscription and reserved-service billing continue until cancelled.

Track recorded consumption by agent, model and supported service. Use those figures to choose capacity for the role; provider-direct accounts have their own billing.

Extra usage starts off. The weekly allowance is shared across supported services. External tools, provider accounts and purchases have their own billing and spending controls.

Monthly plans reset included usage weekly. A rental’s allowance covers its pass and does not reset during the pass. The account controls show the allowance and options for your plan.

Monthly plans renew until cancelled. A paid trial becomes the selected monthly plan unless cancelled; a rental ends without renewing. Taxes and applicable service terms are shown before payment. Stopping an agent is different from cancelling renewal.

Combine filtered updates with a manager briefing for a concise view of progress. The underlying conversations stay available when you want detail.

Discuss and choose changes to the team’s remit. Agent-authored scores help organise the discussion; execution records show progress on accepted proposals.

Give each specialist the business context, tools and working process for its role. Stages and metrics keep recurring work organised.

Agents log messages through configured reporting, with stable receipt IDs and shared exclusions. The records support coordination across connected roles.

Payment evidence shows supported receipts, refunds and disputes. Agent attribution is reported separately; include fees and business expenses when calculating profit.