Agent templates
Each template comes ready with its use cases, capabilities and rules. Before installing it, the admin sees what it turns on; afterwards, they can update or remove it.
Put an agent to work on your systems or ask it what you need, and write with it. Everyone works with their own agents and work accounts, always under the same rules.
Laura opens the chat and sees only the agents her team gave her.
Her home page has the templates her team left ready.
In Ask mode, the agent answers from what it knows about your system, without running anything.
And it cites its source: the rule in the core, with the program and the lines it comes from.
To get work done, Laura picks a template and fills in the detail it asks for.
Before it starts she sees the signed plan and which step waits for a person.
The task starts on her computer, through the Reelease Agent.
The step that changes data stops and waits for approval.
María sees it in her actions inbox, with the task’s data, and approves it.
The task finishes and the result comes back to Laura’s chat.
Your team works with the agent. You decide what gets approved.
You pick the mode in each conversation. Rules, access and the run record are the same in both.
The agent plans the task and runs it in an isolated folder, in the cloud or on the operator’s computer, only with the Sealed Plan and what its rules allow. It starts from an executable use case or, if you allow it, in Goal mode.
Query what the agent knows and draft documents with it, without running anything on your systems. Use it to understand a system and prepare the work.
The list comes from your system, with personal data masked. The agent only sees the row you choose, and choosing doesn't authorize any change.
Agent templates you install in one click and tasks that bring in the right person when needed.
Each template comes ready with its use cases, capabilities and rules. Before installing it, the admin sees what it turns on; afterwards, they can update or remove it.
When the agent needs someone, it opens a task for them (approve, choose or fill in a form) with an owner and a deadline. Your answer is checked against the step’s fields before it carries on. If the deadline passes, the task expires and can no longer be approved.
Everyone works under their own identity and only with what the team has published. Their role decides what they can launch, approve or stop.
Every task runs with the work accounts of the person who started it, stored encrypted and never shared.
By default, every step that changes data waits for someone else, and whoever launched the task can only reject it. The approver sees the exact values first. An admin can change the rule or pre-authorize an action, and it’s recorded.
People on the emergency list can resolve other people’s tasks, never their own, by giving a reason.
The run record shows who started each task, who approved each step and what was done on each system.
How a change goes through
A person starts the task. With Two-Key Approval, by default, they don’t approve their own changes.
Before each change, your rules check it and decide who approves it.
Above your cap: If the amount exceeds the system’s cap, two people approve it or it’s rejected, as you configure.
They see the exact values of the change. If those change, their approval no longer counts.
The notice lands in their inbox and, if they’re not on the platform, by email too. Before the deadline, a reminder.
If no one answers: When the deadline passes, the approval expires and the change doesn’t go out.
Who started it, which rule applied, who approved and the fingerprint of the values.
In Work, the agent plans and runs real tasks on your systems, within its rules. In Ask, it looks up what it knows and drafts documents without running anything.
Anyone with access to the agent: operators from the chat and members from the platform. Every task runs with the accounts of the person who started it, and nobody can run what the rules don’t allow.
That’s the usual path. If the agent’s rules allow it, you can also use Goal mode: you set a goal and the agent builds the plan only with what it has been approved to do. If the plan changes data and has more than two phases, you approve it in full before it starts, and you approve every change with its details in front of you.
Anyone with platform access can cancel the run or suspend the agent: whatever is running stops and nothing else starts until an admin resumes it. What was already done isn’t undone on its own.
The Reelease Agent, for Mac, Windows or Linux, and only when the task needs the operator’s computer. It is installed with a single-use link and only runs Sealed Plans, signed by the platform.
We’ll show you Reelease on one of your cases: what it understands, the plan an agent would propose and what you’d sign off on.
You choose the cookies
We use technical storage to remember your choice and, only if you accept, analytics cookies (Google Analytics) to understand how the site is used. You can change your choice at any time. Learn more in our cookie policy.