A hail damage report for a single address: what fell, when it fell and how hard. Measured on the ground where a Tessera stands, ground-trained by Hailios Fusion AI where one doesn’t, and labeled which is which. A report an adjuster can put in the file and a contractor can stand behind.
THIS ADDRESS · THIS DATE · THIS SEVERITYNO HAIL IS DATA, TOO
Mosaic Reports · event evidencePowered by Hailios Fusion AI · Patent pending
Event report · page one
Specimen
Commercial property · El Paso County, CO
14 May · first impact 22:38
Hail confirmed
Above threshold · this address
Impact count
Measured
Impact energy
Measured
Impact concentration
Measured
Duration
22:38 – 22:52
Max stone size
Measured floor
Most likely size
Modeled · labeled
Confidence
High
Every field says how it was known. No hail is data, too: that report reads the same way.Specimen · field names only
What makes this one different
There might already be a hail report in the file. This is the one built on measurement.
The first question to ask of any hail report: what does it actually know, and what is it guessing at?
Everything a Mosaic report says about the moment of impact is either measured by an instrument at the property or ground-trained by Hailios Fusion AI, a model trained on years of Tessera measurements and still training. It doesn’t need a sensor in the same storm to get it right, and the report says which is which, line by line. A radar-only report can’t make that distinction: it has nothing on the ground to make it against.
01
It is about your property, not the storm’s footprint
Most hail reports are a storm footprint with an address dropped on top. This one starts at your property, whether that’s one roof or a site drawn across many square kilometers, and reports what happened inside it.
02
It can say no hail fell
Radar can’t return a clean negative; it can only say the storm was probably nearby. Mosaic can say no and stand behind it: measured where a Tessera stands, ground-trained everywhere else. No hail is data, too.
Field by field, measured or ground-trained, with the confidence stated in plain language on page one. Nothing is quietly upgraded from estimate to fact.
It states stone size as a floor, then the most likely size
A measured floor where a sensor stands, never inflated, and beside it the size the storm most likely reached, labeled as modeled. Both survive a reinspection; neither is passed off as the other.
RADAR INFERS. WE MEASURE. THAT SENTENCE IS THE ENTIRE DIFFERENCE BETWEEN TWO DOCUMENTS THAT LOOK IDENTICAL FROM ACROSS THE ROOM.
The unit of work
Your property is the unit of work
Radar describes a storm in pixels, and one pixel can cover a whole neighborhood: every roof inside it gets the same answer. A Mosaic report starts from your property instead, one roof or a site outlined across many square kilometers, and drills in closer than a radar pixel alone: measured where a Tessera stands, ground-trained by Hailios Fusion AI where one doesn’t, and labeled which is which.
Every property gets its own answer, and every answer says how it was known.
Report anatomy
What’s in a Post-Event Report
A Post-Event Report answers one question: what happened at this property during this event. Page one says what was measured, what was ground-trained and how sure we are, in plain language, before anything else. These are the fields; the values are your event’s.
Address
The property itself, one roof or a site’s drawn outline, rather than a ZIP code or a county.
Event window
The date, with the time the first impact landed and the last one stopped.
Map and track
Where the storm came from, how it crossed the property and where it went.
Impact count
Where a Tessera stands, how many stones struck it over that window: the raw tally, before anything is interpreted.
Impact energy
Kinetic energy delivered per impact and across the event: the reading that tracks damage most closely.
Impact concentration
How densely those impacts fell — the difference between a scattering of stones and a barrage.
Duration
How long the property spent under falling ice. Two events with the same peak are not the same claim.
Maximum diameter
With a sensor on site, the largest stone it caught: a measured floor, never inflated. Without one, ground-trained by Hailios Fusion AI and labeled so.
Most likely diameter
Modeled outward from that floor using every data point around the site, and labeled as modeled.
Measured vs ground-trained
Which figures come from an instrument on the property and which from Hailios Fusion AI, trained on years of Tessera measurements, labeled field by field.
Evidence
What the finding rests on, stated plainly, and nothing hidden when it points the other way.
Confidence
How sure we are, in plain language, on page one — not buried in an appendix nobody opens.
Status
Preliminary as soon as the event is detected, then final.
Format
A PDF for the file, and structured data with the same values for your systems.
EVERY FIELD SAYS HOW IT WAS KNOWN: MEASURED, GROUND-TRAINED OR MODELED.
On stone size
Maximum stone size is a floor, not a guess.
AT LEAST THIS BIG: MEASURED. MOST LIKELY THIS BIG: MODELED. NEVER BLURRED TOGETHER.
Stone size is the number everyone reaches for and the hardest one to defend. Radar derives it indirectly and publishes a point estimate with no accuracy statement attached, which is why two vendors can hand an adjuster two different sizes for the same storm and neither can be falsified.
We start somewhere else: with measurement. With a Tessera on site, the floor is the largest stone it caught, and it is never inflated: stones of at least this diameter fell here. Across our multi-sensor sites, neighboring sensors differ by 15% on average, so the floor sits close to the truth. Without a sensor on site, the maximum is ground-trained by Hailios Fusion AI, trained on years of Tessera measurements, and the report labels it that way, field by field.
From there the report models outward, using every data point around the site, and states the most likely size beside the measured one, with its certainty, labeled as modeled. You get the number that survives a file review, a reinspection and a dispute, and the best estimate of what the storm actually did, without one being passed off as the other.
Impact energy and impact concentration sit alongside both: measured where a Tessera stands, ground-trained where one doesn’t, and they’re what correlates with damage to a roof, a panel or a car.
The answer radar can’t give
“Nothing above threshold fell here that day” is a finding, not a failure.
77% fewer false positives than radar models · scored against the ground
Radar can’t say no. It can only say the storm was probably somewhere nearby, which is what it also says when nothing happened. Where a Tessera stands, the no is measured. Everywhere else, Hailios Fusion AI answers it, trained on years of ground measurements that include every time nothing fell. Scored against the ground, it reports 77% fewer false positives than radar models: 77% fewer times it says hail fell where none did. That gap is where a great deal of money leaks.
A no-hail reading protects carriers from claims for damage that predates the storm or never happened. It protects honest contractors from driving three hours to a roof that was never hit, and from putting their name to a claim that will not hold. And it lets a site team skip an inspection they never needed.
So a no-hail result is a full report, delivered like any other, with the same confidence statement on page one. You are not paying for a yes. You are paying to stop arguing.
NO HAIL IS DATA, TOO. NOTHING ABOVE THRESHOLD FELL AT THIS ADDRESS ON THAT DATE. STILL EVIDENCE. STILL FILED.
Who it is for
Four people, one document
A hail report gets read by more than one person, and each of them asks it a different question. It has to answer all four without being rewritten for any of them.
Adjusters
“Was there a covered peril at this risk on the date of loss?”
A document that goes in the file and holds up when the file is reviewed.
Contractors, restoration & PDR
“Is this roof worth a claim, or worth a conversation?”
Evidence before the ladder comes off the truck — and before a reputation is attached to it.
Claims managers
“Which of these notifications are real?”
The same report across a book of losses, so triage stops being a coin toss.
Site & asset owners
“Do we send a crew out, and where first?”
What hit each site, from a single roof to a utility-scale solar farm, measured where a Tessera stands, so the inspection goes where the damage is.
Coverage
Measured where we are. Ground-trained where we aren’t.
Where a Tessera stands, the report says measured. Everywhere else, Hailios Fusion AI answers, trained on years of Tessera measurements, and the report says ground-trained, with its confidence stated on page one. The label is never quietly dropped. Europe and Australia have years of measured record too, and training Fusion AI on it is the next step.
MeasuredGround-trained
Powered by Hailios Fusion AI
Post-Event Report · Hail Check · Event Monitoring
Any address in the US, ground-trained by Hailios Fusion AI and measured where a Tessera stands. Outside the US, any site with a Tessera on it, in 150+ countries.
Swath Report
Any US address
MEASURED WHERE A TESSERA STANDS · GROUND-TRAINED BY HAILIOS FUSION AI EVERYWHERE ELSE · LABELED FIELD BY FIELD.
What you can order
Three reports, and monitoring that writes them for you
Order a single report when a single question needs answering, or monitor a property and every Post-Event Report arrives on its own. Every report names the weather data it rests on.
Post-Event Report
One property, one event
Any US address · any Tessera site in 150+ countries
What happened at this property during this event? Start, course and end, stone size with its certainty, the evidence and a map.
Swath Report
One swath, up to 50 km across
Any US address
Where did the hail swath go? An area map of the hail’s extent, in size classes.
Hail Check
One location, one day
Any US address · any Tessera site in 150+ countries
Does this case need a closer look? A finding for one location and one day, with a size class and the kind of evidence behind it.
Event Monitoring
One property, 12 months
Any US address · any Tessera site in 150+ countries
Will I know, and do I get the evidence automatically? Event detection, notifications by email and webhook at the hail class you choose, from H1 to H11, and every Post-Event Report and Hail Check for the property. Monitored properties need no separate Post-Event Reports.
A Tessera on site comes with Event Monitoring for its property.
Years back: should this address have been priced this way?
Hail risk drawn from hail that actually fell, in the US, Canada, Europe, Australia and Japan, with Hailios Fusion AI across the US. Reports tells you what happened; Risk tells you how often it’s likely to.
Tell us what you handle, a claim, a roof or a whole site, and we’ll send a report to read end to end, including how a no-hail result is written where one applies.
