Intelligence & context layer
Meet Haya, Nanoheal’s AI and analytics.
Haya reads what your devices report against what your organisation knows, finds what is going wrong and why, and shows the fix. Your team approves what runs.
66.2DEX score
- 65.1Technology
- 53.8Endpoint
- 76.3Network
- 67.4Sentiment
Score trend
Who is affected
46of 200 machines
below the usable bar this month
- Watches
Each device, site and group against its own history, so a slow drift shows before the calls start.
- Forecasts
Where a metric is heading, with a likely range, so a capacity problem arrives with a date.
- Explains
Which sites and groups drove a change, and how many devices it touches.
- Answers
A question in plain English, with the same number the dashboard shows.
- Suggests
The fix that suits this device and this situation, for your team to approve.
Two identical crashes can need different fixes.
One is a known driver conflict on a build you are retiring. The other is a finance laptop three days from an audit. Haya shows the right fix for each.
Same crash, two incidents
Catches it before the employee does.
Some problems announce themselves. Others build for days. Haya watches for both.
- Its own baseline
- Each device, site and group is compared with its own history, not a fleet average.
- Forecasts
- Metrics are projected forward with a likely range, so a capacity problem arrives with a date attached.
- Less noise
- Related alerts are grouped into one incident, so a single odd spike does not page anyone.
Forecasting and anomaly detection point at a problem. They never act on it alone.
What Haya checks before it suggests a fix.
Four sources, read against each other. Leave one out and the wrong fix can run.
Failure volumeWhole window
Application errors1,899
Application hangs1,502
Unexpected shutdowns622
Bugchecks473
Devices affected44 / 200
Failure kindsRanked by volume
DEX analytics
Performance, crashes, drift and availability across the fleet
A one-off looks like a pattern, and a pattern like a one-off
IT knowledge and SOPs
How your organisation actually resolves this, including the steps nobody wrote down
Every fix is generic rather than yours
ITSM history
What was tried before, and whether it held
Resolutions that already failed get repeated
CMDB and policy
Owner, role, entitlement, compliance obligation, change freeze
The right action lands on a machine that should not have been touched
It does not pick from a catalogue. It writes the entry.
Matching a problem to an old fix is the easy half. Haya works out which steps the situation needs, in what order and with what settings, and writes that up as a new library entry through language-to-automation authoring. A person approves it once, and from then on every device can use it.
Which fix, with which settings.
A display driver that keeps resetting needs the last known-good version restored, the update held back and the display checked. On another build the version and the package differ. Haya keeps that difference in the library entry, so nobody maintains a branching script for it. The agent’s actions are the same on every device; what your organisation would do lives in the entry, not in code.
Ask in plain English. Get the dashboard’s number.
You can ask a question in plain English and get the same number the dashboard shows. Every question, typed or on a dashboard, resolves to the same definition before it is answered, so the two never disagree.
When a question cannot be answered from a defined metric, Haya says so rather than guessing. It runs on your choice of model, including a locally hosted one.
Why do engineering laptops keep freezing this week?
248 display driver resets in 30 days, up 4× on the usual level. 91% are on Windows 11 build 22621, the build being retired, starting after Tuesday’s driver update.
Where it started, and how big it is.
Narrow the search.
Haya gives a root-cause ranking of which sites and device groups drove an anomaly, with what changed where and how many devices are affected.
The dashboards sit on the same definitions as the questions you ask. What the dashboards measure.
248
display driver resets in 30 days
Up 4× on the usual level (dashed)
Starting after Tuesday’s driver update
52 · engineering
Driver rollback
Already in your library
Send for approval, scoped to the 52 devices
Haya suggests the fix. Your team approves what runs.
Flags an anomaly or forecasts a problem.
Decides whether to act, and with which fix.
Answers a question and writes or points to the library entry.
Approves it, then stages and publishes it.
Ranks the sites and groups behind a change.
Chooses where the fix goes first.
Haya itself runs behind approval gates, role-based access, execution traces and PII masking. It does not run anything on devices. Trust & security.
Haya suggests, your team decides, and nothing improvises on your endpoints.