Back to site PEAR
Prompting Solutions · Product

PEAR

Parallel Education Assistance Resource

Teaching material and the evaluations built on it, in one place — where every question is free response, every answer is read and graded against your own reference, and every result adds up to a picture of what your people actually understand.

Scroll to see how it works
Chapter 01

Material and evaluations, in one library

The PEAR library listing lessons and evaluations
The library

Lessons, reference pages and evaluations live side by side, owned by your organization rather than scattered across drives and inboxes. An evaluation is built on top of the material it tests, so the two never drift apart.

Each one carries its own rules: how many questions to ask, whether every category has to be covered, whether the set adapts as the participant works through it.

LIBRARY Lesson Lesson Reference EVALUATION 12 questions Cover every category Adaptive difficulty
An evaluation points back at the material it came from, and carries the rules for how it is administered.
Chapter 02

Free response, not multiple choice

A free-response question with the participant's written answer
A question, answered in the participant's own words

A picked letter tells you almost nothing. PEAR asks people to explain it in their own words — which is what they will actually have to do on the floor, in an audit, or in front of a colleague.

Writing those questions is the part nobody has time for, so PEAR drafts them from your own material and you edit rather than start from a blank page.

YOUR ANSWER
Four options become one open field — and the reasoning behind the answer becomes visible.
Chapter 03

Graded against your reference answer

Every question carries the answer you consider correct. The AI compares what was written against it and returns a verdict — correct, partially correct, or incorrect — together with a written rationale naming exactly what was missing.

Nothing disappears into a black box. The reasoning sits next to the answer, an instructor can override it, and a grade can be re-run when the question or the reference is improved.

WRITTEN ANSWER REFERENCE ANSWER AI grader Correct Partially correct Incorrect + a written rationale for the call
The grade and the reasoning arrive together, and both can be revisited.
Chapter 04

Every question becomes a signal

A single participant's scores broken out by question
One participant, question by question

Because grading happens question by question, and questions belong to categories, a single attempt produces a map instead of a number. You can see which areas someone has solid and which they only half-have.

Repeat the evaluation and the map turns into a trend — evidence that training changed something, rather than a hope that it did.

BY CATEGORY Sampling Documentation Deviations Cleaning OVER TIME attempt 1 → 4
Strengths and gaps by category, and whether they are moving.
Chapter 05

Then zoom out to the whole group

Aggregate scoring across every participant in a cohort
The whole cohort at once

The same data, stacked across everyone who took it. A category the whole cohort is weak in is a training gap. A single question everyone misses is usually a problem with the question or the material, not with the people.

That is the loop PEAR is built around: teach, evaluate, read the result, fix the weakest thing — whichever side of the desk it turns out to be on.

PARTICIPANTS × QUESTIONS Person A Person B Person C Person D Fix the question
When a whole column goes red, the material is the thing that needs work.

That is PEAR.

Teach from your own material, evaluate in your people's own words, and get back something specific enough to act on.

Talk to us See Mantle Back to site