Skip to content

Writing notes in plain sentences

Write what happened the way you would tell a colleague. Relae reads the sentences and pulls out what it needs.

Most donor databases ask you to record a contact as a form. Contact type, outcome code, next action date, stage, dropdown, dropdown, dropdown.

Two things go wrong. Filling it in takes longer than the coffee did, so it does not get done. And the interesting part, that the donor asked about the youth programme and mentioned a parent’s estate, does not fit in any of the boxes, so it is lost.

Relae takes the sentence. You write:

Coffee with Sam Rivera. Asked a lot about the youth programme and mentioned they are redoing their will. Said to come back in the autumn with a proper proposal.

Relae reads that, proposes the follow-up for autumn, notes the programme interest, and flags the estate remark as worth revisiting. You confirm or correct. The record keeps your sentence, because in two years your sentence will be worth more than any structured summary of it.

  • Write it the same day. A note written from memory in November is a note about November.
  • Say what they said. Their words, their concerns, their questions.
  • Say what you committed to. The follow-up is the part that gets lost.
  • Record who else came up. Names are how the relationship graph gets built.
  • Note what you decided not to do. Deciding against an ask this year is worth recording, so next year’s you knows why.

Relae proposes. It does not save anything on its own. Suggested tasks and record updates appear for you to confirm, each showing the phrase it came from, so you can see why it thinks what it thinks.

Nothing is written to a record without you agreeing to it.

Paste a transcript or your notes from a committee meeting or a list review, and Relae separates the record updates from the tasks, and assigns each task to the person named in the room. You confirm the list.

Notes are the most sensitive thing in a donor database. They contain health, family and financial detail that was told to you in confidence.

Before a note is read by AI, the identifiers are taken out: names, emails, phone numbers, addresses. What travels is the shape of the note with placeholders where the people were. The lookup that turns a placeholder back into a name stays inside your workspace and is discarded when the request finishes.

How a note is read with identifiers removedInside your workspace, a note reading "Coffee with Margaret Okonkwo. She asked about the youth programme and mentioned a bequest." has its identifiers replaced, becoming "Coffee with PERSON 1. She asked about the youth programme and mentioned a bequest." Only the replaced version leaves your workspace. The lookup that maps PERSON 1 back to Margaret Okonkwo stays inside the workspace boundary and is discarded when the request finishes. The reply comes back still using PERSON 1, and the real name is put back inside your workspace before you see it.YOUR WORKSPACECoffee withMargaret Okonkwo. She asked about the youthprogramme and mentioned a bequest.identifiers replacedCoffee with[PERSON_1]. She asked about the youth programmeand mentioned a bequest.The lookup stays here[PERSON_1] = Margaret Okonkwoheld for one request, then discardedsentreturnedReads the notesees [PERSON_1]never sees the namenever sees the lookupnothing is keptThe note that travels is the one with the names taken out.Real names are put back inside your workspace, after the reply comes home.

The full account is in how AI reads your notes, including what is never sent and which features are off until you turn them on.