Skip to main content
DoneThat

AI Adoption GuideEducationCredential

Prior Learning Assessment Evaluation

An LLM evaluates portfolio evidence submitted for prior learning assessment credit against defined competency standards and produces a structured credit recommendation.

Education processRecruitAdmitEnrollTeachAssessCredentialGraduateAdvance

By Don, DoneThat’s AI coach · updated

A PLA recommendation cites artifacts and standards, or it stays empty

A prior learning assessment (PLA) evaluation is a quality decision, not a credit posting. The model reads a learner's portfolio against defined competency standards and returns a structured recommendation: which competencies are evidenced, which artifacts support each claim, and whether credit is recommended. If the packet does not contain evidence for a claimed competency, that recommendation stays empty. The model does not invent missing artifacts, does not fill gaps with plausible workplace stories, and does not award credit. Faculty or the registrar awards.

That split matters because PLA sits in a different lane from transfer credit evaluation. Transfer work starts from an official transcript and a sending institution's catalog. PLA starts from evidence the learner produced outside a transcripted course: work products, licenses, military training records, supervised practice, or a structured essay tied to a rubric. A PLA packet dropped into the transfer workflow looks like missing catalog data. A transfer course forced through PLA looks like a portfolio with no artifacts.

The usable output is a recommendation a PLA evaluator can defend in review: competency identifier, standard language, cited artifact (filename, page, timestamp, or rubric row), a short rationale, and a credit suggestion that is either a specific unit or an explicit empty. Empty is a valid result. It is not a soft no that still posts hours.

Load the portfolio packet with the competency rubric

Start with two inputs loaded together, not in sequence as afterthoughts.

The portfolio is the evidence set the institution accepted for this review: files, captions, authenticity statements, and any required cover matrix the learner filled. Authenticity and completeness checks belong to intake (identity, file integrity, required elements present). The evaluation model should not be asked to guess whether a missing signature page exists somewhere else in the student record.

The standards are the competency statements faculty already approved for this PLA pathway: course-level outcomes, program competencies, or a crosswalk the department published. Load the version that governs this review, including effective dates. A recommendation that cites last year's rubric against this year's packet is not usable.

Student information systems and learning platforms (Ellucian, Workday, Anthology, Canvas) hold the learner record, the catalog, assignment dropboxes, and sometimes the PLA request itself. Treat them as the system of record for identity, program, and later posting. Do not treat their document stores as a substitute for the rubric. If the competency list lives in a curriculum map outside those products, attach that map to the evaluation job. If the portfolio lives in a Canvas assignment and the standards live in a program handbook, both go in.

Reject a run that has only one side. A model given a portfolio and no standards will narrate competence. A model given standards and no artifacts will pad with generic professional language.

Map claimed competencies to cited evidence

Work competency by competency, not as a holistic impression of the folder.

For each claimed competency, require a mapping row: the competency ID and exact standard text, the artifact or artifacts cited, the location inside each artifact, and a rationale that a second reader can check without rereading the entire packet. Citation means a pointer a reviewer can open, such as Clinical_log.pdf, entries 14-22, or Capstone_memo.docx, section 3, stakeholder analysis. A sentence that says the portfolio demonstrates leadership, with no pointer, is not a cite.

Related judgment work, such as competency mastery inference from course performance, uses a different evidence type. Course mastery inference reads graded work inside the institution. PLA mapping reads evidence the learner brought from outside or from uncredited practice. Do not collapse them into one score. A strong institutional grade in a related course does not substitute for a missing PLA artifact.

Consider an RN-to-BSN applicant who claims a community-health competency that requires documented population assessment, intervention planning, and evaluation against a published BSN outcome. The packet contains a workplace quality-improvement memo, a state RN license, and a two-page reflection. The license confirms licensure, not that competency. The memo describes a unit-level supply change with no population data. The reflection restates the outcome language. The correct mapping is: license, not evidence for this competency; memo, does not meet the population-assessment standard (cite the memo's scope); reflection, no independent artifact. The credit recommendation for that competency stays empty. A neighboring competency on interprofessional communication might be supported if the memo includes named roles, a decision trail, and an outcome the standard actually asks for. Leave the community-health row empty rather than borrowing the memo.

Watch for invented artifacts. If the learner's cover sheet lists a policy brief that is not in the file set, the model must not reconstruct one from the reflection. If a truncated upload hides a cited page, say the cite cannot be verified. Guessing the missing page is fabrication.

Recommend credit or leave the field empty

The credit field is either a recommendation tied to the institution's PLA unit rules, or it is empty.

Recommend credit only when the mapped evidence meets the standard at the level faculty defined (introduced, practiced, mastered, or the scale the rubric uses). State the recommended units or course equivalent in the institution's terms, and keep the mapping attached. If evidence is partial, do not round up to a full course to be helpful. Partial evidence can be a comment for the evaluator. It does not become posted hours.

Recommend none by leaving credit empty when artifacts are missing, off-standard, unverified, or cited to the wrong competency. Empty is the quality outcome. Inventing a workshop certificate, an hours log, or a typical RN experience paragraph to complete the row is a failure mode, even if the prose sounds professional.

Do not auto-post. Posting unreviewed PLA into the academic record is the failure that makes the rest of the pipeline look successful while it is wrong. Degree progress, financial aid, and later automated degree audit will consume whatever hit the record. Hours on the transcript that nobody signed are a data-quality incident.

Faculty or a designated registrar evaluator awards. The model's job ends at a structured packet: mappings, cites, empty or recommend, and a short list of unverifiable items. Human award is not a rubber stamp on a narrative summary. It is acceptance or rejection of specific rows. The reviewer checks cites, not just the credit number. Post to the student record only after that award, with evaluator identity and date attached.

Keep PLA out of the transfer credit queue

PLA and transfer both can produce resident-equivalent credit, and both can feed a degree audit. They are not the same evaluation.

Transfer evaluation asks whether a cataloged course from another institution matches a receiving course or elective bucket, usually from transcript plus catalog. PLA evaluation asks whether this learner's evidence meets this institution's competency standard, usually from a portfolio plus rubric. Sending a PLA packet through transfer rules produces a broken match: no school code, no course number, no grade. Sending a transfer transcript through PLA produces a portfolio with one artifact (the transcript) that was never meant to be scored as workplace evidence.

Keep queues, reason codes, and posting types distinct in Ellucian, Workday, Anthology, or Canvas-adjacent records so later consumers can tell PLA hours from transfer hours. If awarded PLA later appears on a competency-oriented record, competency transcript generation should inherit the awarded competencies and the human award, not the model's draft narrative.

Is this worth automating for you?

Whether this pays back depends on how much time it takes your team today. Most teams estimate that from memory, and the estimate is usually wrong in one direction or the other.

DoneThat reconstructs where the time actually went, with no timers to forget, so you can measure the baseline before committing to a project and check the gain afterward.

Measure the baseline first