Facility team
Arrive knowing the affected device, likely cause and recent readings - so the visit starts as a repair, not a search.
Catch routine losses before they become portfolio costs.
Volts agents watch energy, water, billing and property workflows, explain each exception and propose the next step. They act only within your permissions and approval rules.
Please enter a valid email address.
How an agent works
From silent exception to owned next step.
An agent does more than raise an alert. It keeps watch, gathers the relevant evidence, explains why the exception matters and prepares the next action for the authorised person.
- Watch the selected sites, meters and workflows on the schedule configured for the agent.
- Show the readings, records and rule that triggered the finding.
- Propose or execute the actions you allow, inside the permissions set by your team.
- Record the finding, decision, action and result in one audit trail.
Energy & water
Stop the four costs buildings hide best.
These agents focus on leakage, peaks, expensive operating hours and loads that never switch off. Each one links the technical signal to the cost line it can change.
- Losses Detector - finds abnormal water or energy loss early, shows the affected meter and proposes the investigation.
- Peak Load Guard - watches the demand window and proposes or executes the response you have approved before the charge is fixed.
- Tariff Optimizer - identifies flexible loads that could move to a lower-cost period.
- Standby Energy Killer - finds sites, zones and equipment that keep consuming outside approved operating hours.
Property & finance
Give month-end back to the people running it.
These agents screen repetitive property and finance work before it becomes a late close, missed renewal or unexplained cost. People keep the decision; the agent prepares the evidence and next step.
- Expense Anomaly Watch - flags unusual or potentially duplicated costs before approval.
- Dunning & Arrears Reminders - prepares tenant follow-up from current balances; a person reviews before anything is sent.
- Lease Lifecycle Watch - surfaces expiring leases, notice dates and unsettled deposits from the lease record.
- Period-Close Assistant - checks missing readings, unresolved exceptions and approval status before finance closes the billing period.
In the workflow
An alert changes nothing. Closed work does.
Every finding can carry its site, device, financial impact and supporting readings into a task with an owner, due date and status. No one has to recreate the context in another system.
- Preserve the location, device, period and evidence behind the finding.
- Turn an approved proposal into work with a named owner and deadline.
- Verify the result against the same readings that exposed the problem.
One finding, three outcomes
Give each stakeholder the decision they own.
The same exception means different work to facilities, property management and the asset owner. Volts keeps one evidence trail while presenting the operational, administrative or financial consequence each role needs.
Property management
See arrears, expense anomalies, lease dates and period-close blockers before they become tenant escalations or late administration.
Asset management
See which exceptions affect owner cost, service charge, tenant risk and the evidence required for portfolio decisions.
Your portfolio
Let each building define what normal means.
A warehouse that runs at night should not look like an office that failed to switch off. Agent thresholds, schedules, sites and permissions follow the real operating pattern of each asset.
- Set thresholds and schedules per site, utility and operating period.
- Start read-only and widen the allowed action scope only when your team chooses.
- Reuse approved configurations when a comparable site joins the portfolio.
The return
Judge every agent by a line you can check.
Measure an agent against the peak charge, tariff spend, standby load, water loss, arrears or time-to-close it was introduced to change - not against a general promise about AI.
- Agree the target line and baseline before the agent starts.
- Keep every detected event, approval and action connected to that measure.
- Report the before-and-after result on the same measure.
Under your control
Autonomy is a permission. Not a promise.
Every agent begins with the scope your team grants. Read-only can remain permanent; sensitive actions can wait for approval; authorised automation stays inside explicit site and equipment limits.
- Define sites, data, action types and approval roles per agent.
- Record each finding, recommendation, approval and executed command.
- Stop command execution from the platform at any time.
Put agents above the systems already working.
Agents use authorised readings and records from the Volts operating layer. Existing meters, BMS and business systems remain the source where they are reliable; Volts adds context, workflow and control.
Works with what you have
Eight pre-built agents
Custom agents
Approvals & audit log
Scope per site
FAQ
AI agents with the boundaries made clear
What exactly do the eight agents do?
Four focus on consumption: Losses Detector, Peak Load Guard, Tariff Optimizer and Standby Energy Killer. Four focus on property and finance: Expense Anomaly Watch, Dunning & Arrears Reminders, Lease Lifecycle Watch and Period-Close Assistant.
Do agents act on their own?
Only inside the scope you grant. An agent can remain read-only, require approval or execute the actions you allow. Every permitted action is recorded.
Can Volts build an agent for our workflow?
Custom agents use the same site scope, permissions, approvals and audit trail as the standard agents.
How do we know an agent is saving money?
Each agent is tied to a defined measure such as peak charge, tariff spend, standby load, water loss, arrears or close time. Results are reported on that measure and retain the actions behind the number.
What happens when an agent gets something wrong?
The finding remains visible with its evidence. An authorised user can reject the proposal, change the rule or threshold and keep the agent read-only while the configuration is reviewed.
What data can an agent access?
Only the sites, devices and records allowed by its configuration and the user’s role.
Can an agent send messages to tenants or suppliers?
Drafting and sending are separate steps, so the recipient always knows when a person approved the message.
Can we test an agent before allowing any action?
Yes. Begin in read-only mode, review the findings against real site data and agree the thresholds and success measure before granting any wider permission.
Put one costly routine under continuous watch.
Bring one recurring exception and the data your team uses to find it today. We will show which agent fits, what evidence it needs and where human approval remains.