How it works

One record, all the way through.

A housing team logs a repair report for Unit 4B. This page follows it from the questions you write to the answers your team works with. It is one of the jobs in the damp, mould and repairs use case.

Record · repair report · Unit 4BREP-2118

Water stain spreading on the ceiling below the bathroom. Outlet next to the sink is warm to the touch.

What you wroteWhat came back · illustration, not a runFollowingUnit 4BWater stain spreading on the ceiling below the bathroom. Outlet next to the sink is warm to the touch.

First, say what a record is made of.

A record can have more than one field. This prism has two, and some context that every question reads first.

Location
location
Text
text
Context · optional

Repair reports from tenants of rented flats. Tenants live in every unit.

Unit 4B arrives as those two fields.

location
Unit 4B, bathroom
text
Water stain spreading on the ceiling below the bathroom. Outlet next to the sink is warm to the touch.

Paste records into the app to try them, send one record per call from your own system with up to about 100,000 characters across its fields, or run a whole CSV, TSV or JSON file of up to 10,000 rows from the app.

Then write the questions, the way you'd brief a colleague.

Edit a question any time. Answers already given stay as they were.

Every answer says how sure it is, and the spread behind it comes back too. It is how sure it is on this text, not a measured accuracy. A close call looks different from an easy one.

Which trade should handle it?

Choice
  • Electricalwiring, outlets, switches, lights
  • Plumbingpipes, leaks, drains, taps
  • Structurewalls, ceilings, floors, stairs
  • HVACheating, cooling, vents
  • Otheralways possible
  • Not enough informationalways possible
  • Electrical71%
  • Plumbing26%close call
  • Structure2%
  • HVAC1%

Is it a safety risk?

Yes / No

Say yes if someone could get hurt before it's fixed.

Yes96%easy one

You decide what counts as sure enough.

How severe is it?

Scale
  1. Minorcosmetic, or a nuisance
  2. Moderategets in the way of using the room
  3. Majordamage is spreading, or a system is down
  4. Criticalsomeone could get hurt now
Major57%

A scale comes back as a spread over its levels, in the order you wrote them.

When none of your options fit, it says so.

Another record, same form

Friday lunch: 12 sandwiches, 3 vegetarian.

It isn't a building issue, so no trade is right. Prismlet doesn't reach for the nearest one.

  • Which trade?Other
  • Safety risk?No 98%
  • How severe?Minor 93%

The other questions still answer.

Unit 3A. Check again next visit.

Which trade?Not enough information

The other outcome you never have to write. It comes back when a record doesn't say enough to decide.

Every record becomes a row.

Each question is a column, and every record is answered the same way. Sort it, filter it, or send each trade its own list, in the tools your team already uses. A developer connects those once; everything above happens in the app.

RecordTradeSafetySeverity
Unit 4BElectrical 71%Yes 96%Major 57%
Friday lunchOtherNo 98%Minor 93%
Stair CStructure 91%Yes 93%Moderate 62%
Roof 2Structure 64%No 88%Minor 71%

For developers: one API call. Read the API docs