Assessment methodology
From workplace capability to evidence-backed interpretation
twoprune uses a fixed six-pillar framework and a governed evidence chain to interpret performance in one realistic, time-bounded assessment context.
The methodology separates what the task was designed to elicit, what was actually observed, how that evidence was interpreted and how much confidence the report should place in the result.
The assessment evidence chain
A defensible report should preserve the path from the capability of interest to the evidence and the final interpretation.
Key points
- 1. Capability
Define the applied AI fluency behaviour of interest through the six fixed pillars.
- 2. Task
Create a realistic assignment with relevant decisions, constraints, materials and an expected deliverable.
- 3. Observation
Collect the submitted work and only the workflow evidence that was actually available and attributable to the session.
- 4. Interpretation
Connect observations to scenario-relevant criteria without converting missing or ambiguous evidence into unsupported conclusions.
- 5. Report
Communicate the result, evidence strength, confidence, limitations, risks and useful follow-up for human review.
The six AI fluency pillars
All six pillars remain part of the universal framework. Scenario design can change which evidence opportunities are most relevant, but it should not erase a pillar or invent ungoverned weights.
Context Framing
How well the participant understands the task, audience, constraints, decision context and success criteria.
Evidence Navigation
How well the participant finds, reviews, weighs and uses relevant information from the available materials.
AI Orchestration
How effectively the participant directs, decomposes, iterates with and critiques AI support.
Verification & Risk Control
How well the participant checks outputs, manages uncertainty, catches unsupported claims and controls operational risk.
Judgment & Synthesis
How well the participant combines evidence, trade-offs and domain reasoning into a defensible recommendation.
Communication & Delivery
How clearly and usefully the participant produces a business deliverable for the intended audience.
Outcome evidence and process evidence
The final deliverable is essential evidence, but it may not explain how the participant arrived at the result. Available process observations can add context about source review, AI use, verification, revisions and time management.
Process telemetry has limits. Opening a document does not prove it was understood, and an event log cannot fully reveal internal reasoning. Interpretation should therefore use traceable observations and state where inference remains uncertain.
Key points
- Submitted outcome
The business deliverable, supporting calculations, citations and explicit assumptions supplied at submission.
- Available workflow evidence
Session-attributable source review, AI interaction, research, revision and timing signals where they were captured.
- Scenario context
The assignment, expected deliverable, evidence opportunities, time limit and relevant interpretation contract.
Confidence, missingness and calibration
Score confidence should reflect the strength, coverage and traceability of the available evidence. If a relevant opportunity was not present or an observation is unavailable, the report should disclose that limitation.
Missing evidence is not negative evidence. A report should not infer that a participant failed to verify an output merely because a verification action was not captured.
Percentiles require a relevant comparison group and comparable assessment conditions. When cohorts are early, small or mismatched, report the comparison as directional or unavailable.
Intended use and human judgment
twoprune provides structured assessment evidence. Organizations may use that evidence within development, enablement, training evaluation or other legitimate human-led workflows.
The platform does not make automated employment decisions, predict future job performance or replace professional review. The strongest use of the report is to understand what the participant demonstrated, what evidence supports that interpretation and what remains uncertain.
Discuss an assessment pilot
Tell us what kind of AI-enabled work you want to understand. We will help determine whether an existing assessment scenario fits the context.
Request early access