aQ Assistant: ask about any site, zone or device in plain language.
aQ Assistant is a multi-agent assistant in a side panel. It knows which site and page you are on, queries the platform's data through tools, and answers with analysis, including a warning when the data behind the answer is incomplete.
From question to analysis.
Context
The site and page you are viewing become the starting point, so "this site" needs no explanation.
Tools
It calls tools that query the Aqueduct database for consumption, demand, events and devices.
Analysis
Results come back as figures, comparisons and observations, not a single number.
Data check
It reports coverage and flags periods with too few readings to compare.
What one electricity summary contained.
Asked for a site's electricity summary and how this week compared with recent weeks, the assistant returned:
Consumption to date
With the share of the month covered and a projected full-month figure.
Average and peak demand
For the period, from the site's meters.
Week-by-week table
With the key observations written out.
A data quality warning
Some days had almost no readings, so the weekly comparison was unreliable.
Why plain-language access matters.
HVACR mechanic openings a year are projected in the US from 2025 to 2035.
Source: US BLSmedian savings from energy information systems in DOE's campaign of about 6,000 buildings, before any control.
Source: US DOE Smart Energy Analytics CampaignAsk your facility a question. Get an answer you can act on.
aQ Assistant knows which site, floor and device you are looking at. Ask in plain language and get consumption, peak demand and trends without walking from the main meter down to each sensor.
- Context aware: it reads the site and page you are on
- Analytical, not a lookup: trends, comparisons and observations
- Honest about data quality, with a warning when coverage is low
Related
Common questions
aQ Assistant responds when someone types a question. Aqueduct agents work in the background, watch telemetry and the twins, and bring problems to you with a diagnosis and a proposed action. Many teams use the assistant to explore what an agent raised. See agentic AI.
Start with a demand question, such as the size of today's peak and the feeder behind it. Then try a zone that raised an air alert, or a device whose events you want explained. The suggested prompts in the panel cover the common openers. See peak demand management.
aQ Assistant answers from your own site data in aQ Core, not from the open web. Every figure in a reply traces back to a specific meter or sensor reading, so an engineer can check it on the device page. See aQ Core.
aQ Assistant says so. It reports how much of the period the meters covered and warns when some days have too few readings for a fair comparison, so nobody acts on a week-over-week change that is really a data gap. See measurement and verification.
See Aqueduct on your own facility.
Bring a site, a utility bill or a question. We will show you where the loss is.