← Use cases/Example · Operations

Operations desk.
Fast answers, within the user’s reach.

An example of what an operations team builds and runs on Xema, not a fixed product: the same foundation takes your data and your rules. An agent reads live public data, such as prices and the weather forecast, and recommends a maintenance window. When it asks to book a field crew, a supervisor approves. The agent can never act beyond what the person it works for may do.

answers from live databookings approved by a supervisorrefusals audited
  1. 01Read live public data
  2. 02Recommend a maintenance window
  3. 03Ask to book a field crew
  4. 04Supervisor approves the booking · a person decides
  5. 05Post to Teams or Slack
  6. 06Refuse what no grant allows
Business outcome

Fast answers, and the agent cannot act beyond what its user may do.

What Xema provides
  • An agent that reads the data and drafts the plan
  • A supervisor approval on the high-risk booking
  • Every refusal in the audit log
01 Outcome

A recommendation in minutes,
an action only when allowed.

A grid operator, a plant or a facilities team has to pick a maintenance window. The answer depends on prices and the weather, which change by the hour. Here an agent gathers the data and proposes the window at once. The step that spends money or moves people waits for a supervisor.

02 When it runs

When someone asks the desk.

An operator asks when the next window should be. The agent works for that person, so everything it does is bounded by what that person may do.

operator asks→window recommended→crew booking approvednothing beyond the user’s reach
03 What it does

Reads. Recommends.
Asks before it acts.

i
Reads live public data
Current prices and the weather forecast, read-only. Reading does not need an approval.
read
ii
Recommends a window
The agent weighs the data and proposes the maintenance window, with the reasons beside it.
a recommendation
iii
Asks to book a field crew
Booking a crew changes something in the real world. It is a high-risk action, so it does not run on the agent’s say-so.
high risk
iv
Waits for the supervisor
The supervisor sees what is being booked and why. The booking runs once, after the answer.
a person decides
v
Tells the team in Teams or Slack
With a grant, the agent posts the booking to the team channel through the connector. A high-risk post can wait for an approval too.
a connector
vi
Refuses what no grant allows
If the agent tries to post somewhere it has no grant for, the call is refused and says why. The refusal is written to the audit log.
refused, and recorded
04 What you get back

Speed you can defend.
Every refusal on record.

The desk answers quickly because the checks are part of the path, not a meeting. You can see what the agent read, what it recommended, who approved the booking, and what it was refused.

The agent is fast, but it is never wider than its user. A person with less authority gets an agent with less authority.

05 Extending

Where teams take it next.

  • A
    More data sources: add another public feed or an internal system. Read maintenance manuals from SharePoint or Google Drive with the operator’s own access. Each source is a registered capability with its own risk tier.
  • B
    Choose what needs a supervisor: reads stay open, writes can wait for one person, and admin actions can need more.
  • C
    Run it on a schedule: have the desk produce the window recommendation every morning, before anyone asks.
  • D
    Hand work to other agents: let the desk ask a purchasing agent for a quote over A2A, with the same decisions and audit on both sides.
  • E
    See how reach is decided: Trust & control explains the tiers, the approvals and the refusals.
Previous: Software deliveryAll use cases →
Operations desk — Xema