Two Prune: AI fluency assessment through realistic work

The assessment is a piece of realistic work

The participant enters a simulated workplace, receives an assignment, reviews evidence, works with available tools and submits a business deliverable. The format is designed to create a consistent opportunity to observe applied AI-enabled work, rather than asking only what someone knows or how confident they feel.

The public Two Prune product description presents eight structured assessment scenarios covering analytical, operational and commercial work. Each scenario has a defined context, time limit, evidence set and expected output. That bounded design makes the conditions of the observation visible.

A simulation is still a simulation. Its evidence belongs to the task that produced it. Organizations should interpret the result alongside the scenario, the material available to the participant and any confidence limitations, rather than treating the exercise as proof of performance in every future setting.

Sources: [1] [2]

Six connected pillars organize the evidence

Two Prune groups observed work into context framing, evidence navigation, AI orchestration, verification and risk control, judgment and synthesis, and communication and delivery. These dimensions help a reviewer locate strengths, development areas and missing evidence without reducing the entire session to an unexplained number.

The pillars follow the work from understanding the assignment through producing a usable result. Context affects which evidence matters; evidence affects what the AI should be asked to do; verification can force a revised judgment; and communication determines whether the final result is useful to its audience.

The report links interpretations to evidence available from the submitted deliverable and recorded workflow. If the assessment did not create a fair opportunity to demonstrate a behavior, or the record cannot support a conclusion, the limitation should remain visible rather than being converted into a negative inference.

Sources: [1] [2]

The report supports a human conversation

The report is intended to help a customer understand what was observed, why the evidence supports the interpretation and where confidence is limited. It can inform capability development, programme evaluation or follow-up questions, while the customer retains responsibility for decisions made outside the assessment.

A useful review distinguishes a participant's final output from the process evidence behind it. A polished document may contain an unsupported claim, while a visible revision may show that an error was found and repaired. Looking at both makes the interpretation more specific.

Two Prune's product boundary matters: it is not an applicant tracker, resume screener or automated decision system. It provides task-bounded assessment evidence for human interpretation. The sample report demonstrates the intended format, including pillar-level observations, evidence strength and calibration language.

• Use the scenario to understand what the participant was asked to do.
• Use evidence links to examine how an interpretation was formed.
• Use confidence notes to identify what the session could not establish.
• Keep consequential decisions with accountable people.

Sources: [1] [2]