Near misses as part of your prevention strategy
Capture near misses
systematically.
Employees describe what happened in thirty seconds – spoken or typed, anonymous, no form. The AI turns it into data you can analyse.
Features
For the workforce
Tell us, don't fill in a form.
A QR code at the machine, the gate or in the vehicle leads straight into the report, without signing in. Speak or type: both end up in the same analysis.
- Three minutes at most of recording, with a timer and a level meter that warns when it is too quiet.
- The audio file is not stored, only the text is processed further.
- Anyone who prefers not to speak can type one or two sentences in their own words.
A rebuild of the reporting view. Runs in the browser, no app to install.
0:47
Recording … 2:13 left
Names are removed automatically. Reports serve prevention, not blame.
Free text becomes structure
The AI suggests, the safety officer decides.
The spoken sentences turn into a complete report: short title, what happened, area, time, activity, equipment, risk potential. Correcting is easier than answering.
- Two follow-up questions at most, answered by tapping tags, instead of a form with eight mandatory fields.
- Names are removed automatically, from every field, including the free text.
- The cause is phrased systemically: not “careless”, but “no fixed place for extension leads”.
The classification is a suggestion. It is reviewed by the safety officer.
2 questions left
Were there any particular conditions?
Multiple answers possible
Please check briefly
Everything important is filled in already. Only change what is wrong.
Short title
What happened?
Area
Location
When
Risk potential
Detected automatically · Electrical safety
No fixed place for extension leads and power strips; damaged cables only surface at the annual inspection.
This classification is reviewed by the safety officer.
For the safety officer
The loop closes, even with a no.
A measure is created for a report or a whole pattern, with an owner, a deadline and a status. Feedback to the people who reported is a field of its own.
- One sentence in plain language, separate from the technical description. That is exactly what the people who reported read.
- One measure answers everyone who reported the same thing, without anyone becoming identifiable.
- Even a no comes with a reason. Silence is the most common reason people never report again.
Status planned → implemented → effectiveness verified, each with owner and deadline.
Measure
Whatever is decided here is seen by everyone who reported this pattern, without revealing their identity.
Fixed cable reels above the work stations
geplantA ceiling reel is mounted above every work station so no extension lead runs across the floor.
Workshop management · due 31 Oct 2026 · applies to workshop / vehicles
Feedback to the people who reported
One sentence in plain language. This is the text people who reported get to read.
No measure · 15 Sep 2026
The area is being redesigned during the autumn refit anyway; an interim fix would create more traffic.
Single cases become patterns
From three reports sharing a cause.
Below that it is a single case, and that is exactly what it says. Every pattern card names the focus area, the most common condition and the share at high risk, and it expands down to each report.
- Pattern detection is arithmetic, not AI. Traceable and expandable at any time.
- The cause comes from the free text, not from a drop-down chosen by the person reporting.
- The word field shows terms across every account of what happened: cable, grating, walkway.
A rising curve first shows a greater willingness to report, not more danger.
Slips and trips
8 of the 31 reports trace back to slips and trips, 5 of them in the treatment plant (63 %), mostly under the condition “wet”, 2 of them with high risk potential.
Show evidence (8)
Cause categories
Derived automatically from the free-text account, not picked by the person reporting.
Terms from the free texts
The number says in how many reports the term appears.
Why people report again
Reporting has to pay off.
Not through bonuses, but because it becomes visible that something changed thanks to your own report. Right after submitting, and again later once something has come of it.
- The progress trail across six stages also shows where things are stuck. An honest interim status keeps trust.
- “This came of it: …”: the safety officer’s sentence reaches the person who reported, without revealing who they are.
- No leaderboards, no points for individuals. Reporting and implementing are rewarded, never a low accident count.
Every figure comes from real data in the application, nothing is estimated. Anyone reporting via the QR code without signing in sees this feedback right after submitting; the later progress is visible once signed in.
Thank you, this will be acted on!
3 reports about Electrical safety, which makes it a pattern.
Stored anonymously.
What happens next
Received15.09.2026
Seen by the safety officer
Recognised as a pattern
3 similar reportsDecision pending
Implemented
Effectiveness verified
What came of your report
Every workshop station now has a ceiling reel. No cable lies in the walkway any more.
Your report was one of 3 that led to this change.No months-long rollout. tutelaris is ready straight away.
No IT project, no weeks of onboarding: print the QR codes, put them up, report. For the pilot we set up your areas together.
Reported in thirty seconds
A voice message or a few sentences, in your own words. No form, no account – scan the QR code and report.
The AI turns it into analysable data
Area, activity, cause and risk are derived from what was said, plus at most two follow-up questions.
Impact becomes visible
After submitting, the reporter sees whether the report belongs to a pattern and follows it across six stages. No points, no leaderboard.
Anonymous from the start
Names are removed from every field automatically. A pseudonym is stored instead.
Built for the shop floor – from the shift to the safety officer.
One app for all three roles: employees report, the safety officer reviews and decides, managers see what changed.
Report without signing in, without a form.
QR codes at machines, gates or vehicles lead straight into the report – spoken or typed, in your own words.
Hall 3 · QR code
The AI suggests, the human decides
The AI proposes a classification and a risk level. The safety officer reviews and decides – pattern detection itself is arithmetic, not AI.
Keep Your First-Aid Log Digital
Every first-aid treatment is recorded digitally on-site in the log – fully documented, archived in DGUV-compliant form, and accessible at any time.
Exportable to any format
A single report or a filtered set as PDF, CSV, JSON or Excel – with the footer “Anonymised · Serves prevention, not blame”.
Why Companies Choose tutelaris
Ask instead of search
The safety officer can query their own body of reports: are there comparable cases? Which measure works here? Every answer links to the reports it came from.
Self-efficacy instead of chasing points
People who see that their report led to a change report again. Hence team goals per area, streaks that only go up – and no individual leaderboards.
Analysable instead of filed away
The cause is derived from the free text, not picked by the reporter. That is what makes analysis possible where PDF forms fail.
Become one of our first five pilot sites.
Three months free, one or two areas are enough. In return we ask for honest feedback from day-to-day work.