Under construction. You’re seeing SiteRisk before it’s finished.

Weather intelligence for the people who have to make the call.

Every decision is a risk. SiteRisk makes sure it’s not a guess.

We use hundreds of forecasts to generate insights for every weather hazard against your specific needs, site‑by‑site. We also tell you how much to trust our forecasts, because no two forecasts are alike.

We’re taking on a small number of pilot sites this season. For additional questions, email dalton@siterisk.ai.

Thursday’s calls, site by site

Thursday, October 1, 2026
  1. Timber Ridge Roof

    WindCaution

    12a6a12p6p12a

    Peak 50–60 mph around 1 PM. Secure loose items by noon.

    Confidence Moderate

  2. Potrillo Ln Framing

    LightningHeads-up

    12a6a12p6p12a

    Peak activity 2–3 PM. Plan for delays between 2 and 3 PM.

    Confidence High

  3. Gaylord Texan - Landscaping

    Clear

    12a6a12p6p12a

    Confidence High

Every project, every hazard, one screen. See what needs your attention at a glance, then open any site for the full picture.

01 We keep score

Every call gets checked after the storm passes.

Most forecast verification checks whether a forecast matched the atmosphere. Ours checks whether it would have gotten your crew to the right call, at your threshold, before your call time.

Graded against independent storm reports and radar data, not our own model checking itself.

A claim isn’t proof. This is.

Verification
Windgusts over 40 mph
34Hits
2Misses
5False alarms
Raintotals over 0.5″
28Hits
4Misses
3False alarms
Hailany hail at the site
11Hits
3Misses
9False alarms

When we’re not sure, we say so.

Two forecasts can carry the same headline number and mean completely different days. One is worth mobilizing on. The other is worth checking again at noon. You should be able to tell which one you’re holding without being a meteorologist.

MOST LIKELY 28 MPH — BOTH DAYS MODELS AGREE 26 31 a five mile-an-hour question MODELS DISAGREE 18 44 a breezy day, or a stand-down 20 28 36 44 mph
Same most-likely number, same site, same hour. The top one you can plan around. The bottom one you cannot — and the board tells you which is which before you send anybody out.

Confidence isn’t one measurement. Near-term it’s how tightly the runs agree and whether independent models tell the same story. Further out those go quiet, and what’s left is how unusual the day looks against twenty years of local history. The mix shifts as the horizon stretches — which is why the board draws near-term windows solid and gives you a dashed watch further out, instead of a made-up number.

We compute. You decide.

Worth being plain about, because you have been pitched “AI” four times this year: nothing on this board decides anything.

What SiteRisk does

  • Reads every model run, every cycle
  • Grades each condition at every level that matters
  • Measures it against your site, your limits, your hours
  • Says how likely, and how sure

What you do

  • Make the call

Every number here is computed from the weather models themselves. No risk score, no 1-to-10, no language model’s opinion. You bring the judgment; we make sure you aren’t guessing at the arithmetic.