Skip to content
Nonprofit Solutions AIMission-first AI field lab

Nonprofit AI field kit

Browser-safe method

Pilot evidence log

A bounded experiment needs a visible record of quality, burden, risk, and decisions.

Intended output

A reviewable record that separates observations from conclusions and keeps failures useful.

  1. 01

    Log the case ID, workflow version, reviewer, date, result, time, corrections, and escalation.

  2. 02

    Record unsupported claims, omissions, harmful framing, inaccessible output, sensitive content, and near misses.

  3. 03

    Note staff preparation, review, correction, documentation, and support burden.

  4. 04

    At the gate date, summarize evidence for stop, revise, continue, or scale and name the owner.

Monthly field note

Get practical guidance in your inbox.

Monthly field guidance for mission fit, data boundaries, human review, and bounded AI pilots.
Required