Skip to content
Construction risk management software

Find the risk. Close the loop. Prove it.

SALUS is construction risk management software that connects risk assessment, hazard identification, incidents, and corrective actionsβ€”then tracks every risk to verified closure. Every follow-up stays tied to a named owner, current standard, and field record while Rosie, the AI Admin & Risk Assistant inside SALUS, watches the open loops.

You can't fix the risk you can't see β€” and you can't claim the fix you can't prove.

Live board

Open loops Β· 14 active sites

7 loops open
Near-miss, Level P3Reported 3 days ago β€” silence
Inspection findingWritten up, no owner
Corrective actionOpen 12 days, no status
Toolbox talkSame six, recycled
SWP at the new scopeStale β€” 8 years old
Site conditionsNobody's watching the wind
Portfolio trendInvisible
  1. Near-miss reporting climbs when capture costs seconds

  2. Actions owned by a name, not a company

  3. 97% closed within 24 hours β€” provable

  4. Every action traced to its origin

  5. Rosie watches every open loop

The problem

Most programs can report. Fewer can follow through β€” and prove it.

A paper JHA or isolated hazard register captures one moment. The field reports and hears nothing back, the office assigns and can't verify, and every new project makes the risk picture harder to connect across inspections, incidents, actions, and proof.

Open loops. Drifting standards. Risk you can't see.

In the field

Reports go in. Nothing visibly comes back.

  • Hazards and near-misses reported β€” then silence
  • Findings written up with no owner attached
  • The fix happens, but nobody can prove it
  • The same six toolbox talks, whatever the work
  • SWPs that don't match the new site or scope
  • Paper JHAs signed, not read
Across the company

Growth outruns the standards.

  • Corrective actions open for weeks, assigned to companies instead of people
  • No link between the action and the inspection, incident, or meeting that created it
  • Standards deployed unevenly β€” every site running its own version
  • Near-miss data too thin to mean anything
  • Trends by site, trade, or individual β€” invisible
  • β€œWhere is risk lurking?” β€” unanswerable
The solution

Risk and compliance software built for construction safety

SALUS keeps the construction safety loop connected from field signal to assigned control and verified closure. OSHA evidence stays linked to the inspection, incident, or assessment that created it; current standards stay consistent across sites; assigned controls stay tied to named owners; and every verified fix creates audit-ready closure proof. Rosie monitors what remains open and surfaces patterns across the portfolio.

The crew

Reports β€” and keeps reporting

Because capture takes seconds, and the follow-through is visible. Feedback is what keeps the signals coming.

The signals

Risk becomes visible

Near-misses, findings, and incidents β€” captured in seconds, so more of them arrive. One near miss is a report. Ten are a signal.

Rosie

Watches every open loop. Connects every action to its origin. Surfaces the risk before it surfaces you.

By site Β· By trade Β· By sub Β· By individual

The actions

Closed, with proof

Assigned in-flow to a named person with a due date, tracked to photo-proven closure β€” connected to what created them.

The standards

Deployed, customized

SWPs, JHAs, PTPs, and talks β€” current on every site, sized to the crew, the activity, and the day.

…and the standards loop back to the crew

Signals flow up. Standards flow down. Nothing stays open.

In-depth Β· before / after

Same sites. Same crews. A different picture of risk.

Risk control isn't a time figure β€” it's visibility. Here is the same 14-site quarter, twice: once with reporting that costs 20 minutes, and once with capture that costs 90 seconds. And because the modules are connected, the correlations become readable: form completion against incident rates, training completion against near misses, corrective-action closure against repeat hazards.

Before Β· the lights are off14 sites

Risk is invisible until it's an incident.

The safety pyramid β€” what gets reported

Incidents3

The only layer you see β€” after the fact

Inspection findings11
Near-misses23 logged

The dashed mass never gets reported β€” 20 min of paperwork each

The portfolio β€” where is risk lurking?

?
?
?
?
?
?
!
?
?
?
?
?
?
?

13 unknowns Β· 1 surprise, found at month-end

Open corrective actions β€” age & ownership

19 days Β· β€œthe framing sub”

14 days Β· unassigned

9 days Β· status unknown

47 open Β· avg 19 days Β· proof: take their word

After Β· with SALUSSame 14 sites

Risk is ranked, owned, and closing.

The safety pyramid β€” what gets reported

Incidents1

Fewer β€” because the base finally talks

Inspection findings38
Near-misses96 captured

The signals multiply β€” capture takes 90 seconds

The portfolio β€” where is risk lurking?

A
B+
A
C
B+
A
D
B+
A
C
A
B+
A
B+

Every site graded Β· the D flagged by Rosie β€” visited tomorrow, not month-end

Open corrective actions β€” age & ownership

22 min Β· M. Chen Β· photo βœ“

1 day Β· R. Alvarez Β· photo βœ“

2 days Β· D. Okafor Β· in progress

6 open Β· 97% closed within 24 hrs Β· photo-proven

Lagging indicators become leading ones. The ratios are what the field reports.

Recognize the open loops? They close β€” with names, timestamps, and proof.

See the loop close on your program
04Control Β· One loop

See it. Shut it. Learn from it.

Control 01 Β· Deploy the standardsEvery site its own versionCurrent everywhere, sized to the work

Standards that scale β€” without flattening into boilerplate.

Growth used to force the choice: one generic program everyone ignores, or a tailored one nobody can maintain. Rosie maintains the tailored one β€” SWPs, JHAs, PTPs, and talks deployed to every site, customized to the crew, the activity, and the day, and kept current as the work changes.

01

The SWP, current on every site

Rosie cross-references near-misses against the procedure, flags the gaps β€” including the document nobody has touched in eight years β€” and regenerates it. Update once; every site runs the same current version.

02

The JHA, sized to the task

Conditional logic branches the form to today's work β€” 24 questions become 8 relevant ones. Crews read what they sign, and the hazard data that comes back actually means something.

A Daily Site Safety Report branches from a 24-question full form down to 8 relevant questions when conditional logic hides what doesn't apply
03

The talk, written for this crew, today

Rosie drafts each briefing from the site's own near-misses, the scope, and the conditions β€” hyper-relevant beats recycled, and the QR sign-in makes attendance a record instead of a clipboard.

⏱ SWP for a new scope Β· days β†’ minutes⏱ One current version Β· every site
Control 02 Β· See everythingThin data, silent fieldThe signals multiply

You can't fix the risk you can't see.

Near-miss data is thin for one reason: reporting used to cost 15–20 minutes of paperwork. Make capture cost seconds and the field starts talking β€” observations, inspections, near-misses β€” the rich intel that risk identification actually requires.

01

The observation, spoken

A foreman talks while walking the slab; Rosie structures it into a signed observation record with photos attached. Inspecting and documenting become the same act.

A foreman speaks while walking β€” Rosie turns the voice note into a structured, signed observation record with photos attached
02

The near-miss, captured where it happens

Three levels underground with zero signal: photos, video, voice β€” auto-saved offline, synced complete when signal returns. The report that used to take 20 minutes at a desk takes 90 seconds at the scene.

A near-miss documented in a subterranean parking level β€” photos, video, and voice captured offline, syncing when signal returns
03

The inspection that acts on the spot

The supervisor spots the hazard, and the corrective action is assigned before the walk ends β€” same workflow, no separate module, no escalation chain. Seeing and assigning become one motion.

A hazard reported in four steps β€” supervisor spots it, the foreman gets the corrective action, the fix goes in, the loop closes with photo proof
⏱ Near-miss report Β· 15–20 min β†’ 90 sec⏱ Field intel Β· signals multiplied
Control 03 Β· Close with a nameOpen for weeks, owned by nobodyClosed in hours, with proof

From construction risk assessment to verified corrective action

Actions assigned to "the framing sub" die in an inbox. Actions assigned to Marcus Chen, due 5 PM, with the photo and the location attached, get closed β€” and the closure carries evidence. Every hour an action stays open is open-risk time; closing faster shrinks it, and strengthens OSHA and insurance defense. Every action stays linked to the inspection, incident, form, or meeting that created it, so the loop can be audited end to end.

01

Owned by a name, closed with a photo

The assignment lands on one person's phone with full context and a due date. The fix is photographed, timestamped, and verified β€” this one closed in 22 minutes.

02

Velocity you can manage

47 open actions across 14 sites, average closure two days and improving β€” with overdue items ranked by trade, so the conversation with the framing sub is about their 60%, not a feeling.

Actions overview β€” 47 open corrective actions across 14 sites, 2-day average closure velocity, overdue items ranked by trade
03

Traced to its origin

Every action carries its lineage: the inspection that found it, the JHA it relates to, the evidence that closed it, and the toolbox talk it fed afterward. Nothing floats free of its context.

Corrective Action #1284 β€” full lineage

Closed
Origin
Safety audit β€” Level 3 East walk, 9:41 AM
Finding
Missing Class A fire extinguisher Β· Station #L3-E4
Assigned
Marcus Chen (Site Supervisor) Β· due 5:00 PM
Evidence
Resolution photo Β· 10:03 AM Β· GPS-tagged
Closed
22 minutes after assignment Β· verified
Feeds
Tomorrow's toolbox talk + the portfolio trend line

Nothing floats free of its context

⏱ Time to close Β· 2–3 weeks β†’ 24–48 hrs⏱ 97% within 24 hours β€” provable
Control 04 Β· Rosie watchesTrends invisible until the incidentRisk surfaced, ranked, answered

Where is risk lurking? Rosie answers with evidence.

No human reads every form across fourteen sites β€” Rosie does, continuously. She monitors open actions and trends by site, trade, sub, project, and individual; watches each site's own conditions β€” wind climbing past 38 mph at the tower, a scaffold inspection due at the harbor; flags the anomalies that precede incidents; and tells you where to spend tomorrow.

01

The silence, flagged

Four active sites have zero hazard-assessment submissions this week β€” two of them with 71 workers on site. The most dangerous signal is the one that stopped arriving; Rosie notices the absence.

02

The drift, caught early

A crew's form completion quietly drops 18 points below baseline after a foreman change. A project's incident volume is rising while its reporting quality is falling. Neither is a dashboard you checked β€” they're the compound signals Rosie raises before they become recordables.

Rosie flags an unusual drop in forms completion β€” 61%, 18 points below baseline β€” generated in 2 minutes
03

The response, recommended

Watching closes into acting. When corrective-action closure drifts, Rosie says so β€” and recommends the fix: a safe work procedure change, a toolbox talk, a conversation. When hand-injury near misses spike across two projects, she routes a three-minute refresher to the affected crews before the next morning's hazard assessment. The trend line moves before the incident report does.

Rosie β€” recommended response

Routed
Signal
Hand-injury near misses ↑ Β· two projects
Cross-check
Low scores Β· cut-resistant glove module
Response
3-minute refresher β†’ the affected crews
Lands
Before tomorrow's daily hazard assessment

From calendar event to risk response

04

Tomorrow, decided by the data

β€œWhich site should I go to tomorrow?” Rosie ranks the portfolio β€” open incidents, unassigned actions, open high-risk permits, expiring certs β€” and names the site that needs you, with the risk profile attached.

⏱ Conditions · watched on every site⏱ Trends · by site, trade, sub, project, individual
Who gets what back

Three desks, watching the same loop close.

Safety leaders

Gets back Β· Risk, ranked

Field intel multiplied, trends by site and trade, and a daily answer to β€œwhere do I spend my time?” Field presence goes where the risk is β€” not where the rotation says.

Operations leaders

Gets back Β· Standards that hold at scale

Every site on the same current program, customized to the work. Findings become assigned actions before the walk ends β€” fewer surprises reaching the schedule.

Executives

Gets back Β· Provable control

Closure velocity as a leading indicator underwriters and owners can verify on a live dashboard, the blindside project caught early, and growth that doesn't dilute the program β€” the renewal story itself lives in Proof on Demand.

Both halves of control

More seen. Nothing left open.

The halves reinforce each other: easy capture multiplies the signals, the signals direct the fixes, and proven fixes teach the program β€” so the next standard ships smarter than the last incident.

What you see

  • Near-misses multiply β€” capture that costs seconds
  • Findings documented during the walk, not after
  • Trends by site, trade, project, and individual
  • The anomaly flagged before the incident
  • The site that needs you tomorrow, named

What you close

  • Actions owned by named individuals, due-dated
  • Closure in hours, not weeks β€” photo-proven
  • Every loop traced to its origin, auditable end to end
  • Standards current and customized on every site
  • β€œWhere is risk lurking?” β€” answered with evidence
Risk control at scale

Risk can't hide in a closed loop.

Bring your open corrective-action list and last month's near-misses to a 15-minute demo β€” and watch the loop close with names, timestamps, and proof.

Deploy Β· Capture Β· Assign Β· Close Β· Learn