Build direct measures that still reveal student judgment when AI is available.
The Sprint helps a program strengthen one of the hardest parts of assurance of learning in an AI-enabled environment: deciding what can be inferred from student work when AI may shape the final artifact. The engagement produces a pilot-ready measure, an evidence architecture, and a complete plan for scoring and improvement.
A polished final product can hide the reasoning a program needs to evaluate.
AI can help a student produce a credible memo, analysis, presentation, recommendation, or campaign. Programs still need direct measures that support a defensible inference about the student’s own capability.
What the final artifact may leave unclear
- Who framed the problem and set the direction.
- Which AI outputs the student accepted, rejected, or revised.
- Whether important claims, calculations, or sources were verified.
- How the student handled uncertainty or a flawed recommendation.
- Whether the student can explain the reasoning without relying on the tool.
What an evidence-rich measure can reveal
- The student’s role in shaping the work.
- Key workflow and decision points.
- Verification and revision behavior.
- Professional judgment under the chosen AI condition.
- The ability to explain, adapt, and defend consequential choices.
The goal is enough evidence to interpret learning with confidence. The Sprint uses bounded process evidence and purposeful defense rather than turning assessment into continuous monitoring or surveillance.
Four practices make AI-assisted reasoning more interpretable.
Disclose, document, verify, and defend provide a practical structure for making relevant parts of the student process visible. A measure can use all four practices or a carefully chosen subset based on the competency and the stakes of the decision.
Disclose
Explain the role AI played, why it was used, how it influenced the work, and where final accountability sits.
Document
Preserve enough of the workflow to show how the task developed and where consequential decisions occurred.
Verify
Check claims, sources, logic, calculations, assumptions, and fit when the learning outcome requires reliable professional judgment.
Defend
Explain, adapt, or stand behind the final judgment through a brief written, oral, or applied defense appropriate to the measure.
Design backward from the inference the program needs to make.
The Sprint begins with the student capability the program wants to evaluate. The task, AI condition, evidence requirements, rubric, scoring process, and reassessment plan are then built around that inference.
That sequence matters because adding disclosure language or a process log after an assignment is already designed can create more paperwork without creating better evidence. The measure should collect only the information that helps a scorer interpret the intended competency.
For some outcomes, students may need to work without AI. Other outcomes are better tested with AI available because professional capability includes using the tool well. The Sprint makes that decision explicit and aligns the evidence to the condition.
Move one priority measure through a complete evidence cycle.
A bounded scope keeps the work usable. The first engagement usually focuses on one competency and one direct measure so the program leaves with a design it can actually pilot, score, discuss, and improve.
Clarify the inference
Define what the program needs to know about student capability. Identify the performance that would support that conclusion and the evidence limitations in the current measure.
Set the AI condition
Decide whether AI should be prohibited, limited, permitted, required, or compared across conditions. The decision follows the learning purpose rather than a general institutional preference.
Build the evidence-rich measure
Redesign the task and rubric so artifact evidence is complemented by the right process, verification, or defense evidence.
Pilot and close the loop
Define scorer guidance, calibration, data collection, findings, improvement decisions, and reassessment so the measure becomes part of a usable assurance cycle.
A pilot-ready assessment and the tools to interpret it.
The Sprint connects design, evidence, scoring, and improvement. Each deliverable has a specific job in the assurance process.
Use the Sprint when the inference matters more than the technology.
The strongest trigger is usually a direct-measure question rather than a general desire to “add AI” to assessment.
Your current measure has become harder to interpret
Students can produce stronger-looking work with AI, and the program needs better visibility into the judgment behind the output.
You are defining an AI-relevant competency
The program needs a direct measure that shows whether students can apply, evaluate, verify, or oversee AI in a professional context.
You need a stronger assurance story
The existing AoL process collects results, but the program wants clearer evidence that the measure remains valid under current student working conditions.
The Sprint can stand alone or follow the AI Capability & Evidence Diagnostic when student learning evidence emerges as the most important gap.
Start with one consequential measure.
The entry engagement is intentionally bounded so a program can test the model before expanding it across a full assurance system.
$5,500–$7,500
For one competency and one priority direct measure.
- Design and evidence architecture
- Rubric and scorer guidance
- Pilot and improvement plan
$8,500–$12,500
For a small set of related measures or a more extensive program-level assurance problem.
- Multiple linked measures
- Calibration support
- Program-level evidence map
Scoped separately
For multi-program work, facilitated faculty teams, or a larger assessment redesign initiative.
- Custom measure count
- Faculty working sessions
- Implementation support
Ranges are indicative and finalized after scope. The Sprint is platform-independent and can work inside an existing assurance-of-learning process.
What assessment leaders usually want to know.
Does every direct measure need to prohibit AI?
No. The AI condition follows the competency and the inference the program needs to make. Some measures may need independent performance. Others become more authentic when students use AI and are held accountable for how they use it.
Do we need a separate AI competency?
No. AI-relevant dimensions can be embedded inside existing program competencies when that better fits the curriculum and the school’s quality process.
Can this fit our current AoL system?
Yes. The usual goal is to strengthen the validity and interpretability of an existing direct measure rather than create a parallel assessment system.
Does the approach require a specific platform or proctoring tool?
No. Technology may help collect particular artifacts, but the evidence architecture is designed independently of any one platform. The program can choose tools that fit its own environment and policies.
Start with one competency and one complete evidence cycle.
A focused sprint can establish a model your program tests before deciding how broadly to apply it.