✦ AI-Powered Oral Assessment

Turn submitted work into demonstrated understanding.

Provenance is the artifact-based oral assessment platform that helps faculty verify learning through reviewable, AI-guided conversations—at scale.

Request a demo →
Authentic
assessment
Learning
insights
Built for
security review
Student Artifact
PDF Ocean_Resilience_Report.pdf
1 / 12
−100%+
Recording 4:32
Ocean Resilience in a Changing Climate
1. Introduction

Rising sea temperatures and ocean acidification are driving measurable changes in marine ecosystems. This study analyzes coral reef health across the Indo-Pacific between 2000 and 2020.

2. Methods

We analyzed data from 58 reef sites using in-situ temperature loggers, pH sensors, and benthic surveys. Trends were assessed using generalized additive models (GAMs).

3. Key Findings
Coral Cover (%) Over Time
1007550250 20002005201020152020 Coral Cover (%) Year
4. Conclusion

Coral cover declined by 37% over the last two decades. Local adaptation strategies show promise, but global emissions reduction remains critical.

1
✦
Provenance Reviewer Provenance AI ▾
✦
Provenance Reviewer04:32
Can you walk me through the key factors that influenced coral cover decline in your study?
S
Student04:45
Yes, the primary drivers were rising sea surface temperatures and ocean acidification. Both stress corals physiologically and make them more vulnerable to bleaching events.
✦
Provenance Reviewer04:58
How did you distinguish the impact of temperature from other stressors in your analysis?
S
Student05:13
We used a GAM to isolate temperature effects while controlling for variables like pH, local pollution, and fishing pressure. Temperature remained a significant predictor across all models.
✦
Provenance Reviewer05:28
What are the implications of your findings for coastal management?
Recording...
Speak clearly. The Provenance Reviewer is listening.
The question has changed
“Did AI write this?”
“Can you explain it?”

AI detection alone isn't enough. Provenance shifts integrity from suspicion to evidence — by inviting students to walk through the work they submit.

Not AI detection. Not misconduct adjudication. Provenance doesn't decide whether a student cheated — it produces structured evidence of whether they can explain their own work.

Three steps from submitted work to reviewable evidence.

HOW IT WORKS
01
Meet students with their work

Every student walks through the exact artifact they submitted — paper, model, dataset, or report — with the Provenance Reviewer.

02
Ask them to walk through it

Questions come from the decisions inside that artifact — not a generic test bank — so genuine understanding shows for itself.

03
Get evidence of real understanding

A rubric-mapped evidence record with AI-assisted recommendations for faculty review — clear, linked to the transcript, and human-decided.

Faculty stay in control

The AI suggests. Faculty decide.

Provenance never assesses on its own. Instructors shape the review and own every academic decision.

Approve before it runs

Instructors review, edit, and approve the questions and evaluation criteria before any student begins.

Traceable by design

Every question traces back to the rubric and to a specific decision in the student's own artifact.

Final judgment is human

After the session, faculty review the evidence and AI-assisted recommendations, then make the final call.

Built for the institution

Evidence of understanding, at the scale of a department — or a campus.

Provenance plugs into the systems you already run and produces clear, evidence-linked records of what each student understands.

Grounded in real work
Every question traces to a decision in the student's own artifact.
Scored to your rubric
Outcomes map to the criteria your faculty already use.
Evidence-linked records
Reviewable transcripts that show how each conclusion was reached.
Fits your stack
Designed for FERPA-aligned deployment, LTI 1.3 launch, and faculty-validated results.

Move integrity from
suspicion to evidence.

See a session run on your own assignment. We'll walk your team through the evidence it produces.

Request a demo Talk to our team