TotalApp Docs

Recall

Bring similar past decisions and their results back, with sources, conflicts and an as-of date.

Recall

Recall answers one question: what happened the last time we were in a situation like this? When you open a record, Enterprise Mind looks through memory for similar past cases and shows the closest few in a card, with the decision that was taken, the result, and a link to the original record.

It is not a general chatbot and it does not answer from outside knowledge. A result appears only if it can point to a record, message, call or document in your own tenant, and only if you are allowed to open that record. If nothing fits well, Recall says so instead of showing a weak match.

You meet Recall in three places: as an automatic card on a record, as the search box on the Recall screen, and as the history behind the evidence on an approval card.

Way to use itWhat you doTypical question
Card on a recordOpen any record. The card appears on its own."Has this happened before?"
SearchType a situation in plain words and press Enter."Late check-in, room not verified"
Evidence on approvalsRead the "Why this suggestion" part of the card."What is this suggestion based on?"
As-of dateMove the date control to look back."What did we know in September?"

When to use Recall

  • Before approving something you have not seen in a while.
  • When a case looks familiar but nobody can remember how it ended.
  • When you inherit a module from a colleague.
  • When a customer says "last time you told me...".
app.totalapp.app/enterprise-mind/recall
Room 214 cleaning not verified before check-in As of Sep 12
DECISIONExtra cleaner added3 weeks ago
hospitality/tasks/T-102 · M. Reed
91%
OUTCOMEStill 40 min late3 weeks ago
hospitality/tasks/T-102 → closed
84%
CONFLICTRule says “ready” once cleanedv3
Policy v3 vs. supervisor note: verification required
58%
Both sources kept
  • Sources. Every result links to the original record, who decided and when. If a result cannot cite a record it is not shown.
  • Conflicts. When two sources disagree, both are shown side by side with a conflict badge instead of one being picked silently.
  • As-of date. Move the date control to see what the system knew on that day.
  • Your roles apply. People only recall what they could already open in the module.

Confidence is not a guarantee

The percentage on a result shows how closely it matches your question, not how likely it is to be right. Read the source before you rely on it.

How results are ranked

Recall ranks candidate cases by how well they fit your question. You can change the order with the filters above the list.

SignalEffect
SimilarityCases with the same type, size and conditions rank higher.
RecencyNewer cases rank higher unless you widen the period. A policy change makes older cases less relevant.
Outcome knownCases with a confirmed result rank above cases that are still open.
Same module and roleCases from your own module and role are preferred over distant ones.

If nothing fits well, Recall says so and shows nothing rather than a weak match.

Worked example: a late check-in

A supervisor opens room 214 an hour before check-in and sees it is not marked verified.

  1. Recall opens a card automatically and lists three similar cases from the last two months.
  2. Two cases show the same pattern: an extra cleaner was added and the room was still 40 minutes late.
  3. One result carries a conflict badge: the policy says a room is ready once cleaned, a supervisor note says verification is required.
  4. The supervisor marks the supervisor note as the current rule, writes a short reason and approves the suggested extra cleaner.
  5. Later the room is ready on time. The outcome is recorded and linked back to this decision.

The next supervisor who meets the same situation sees this case first, with the reason and the result.

Related pages

Memory explains what Recall searches. Decision Ledger shows the same entries as a list. Outcome Learning uses recalled cases to check predictions.