Skip to content

Capability 05 · Equip the field

Field Training

Build learning and practice materials from the exact approved source set, with page- or slide-level excerpts attached to generated claims.

The problem

Training drafts can drift from their source material, omit safety context, or silently mix old documents into a new request. Review becomes slow when a learner, facilitator, or MLR reviewer cannot trace a message to an exact passage.

Objective

Turn one or more supplied, approved PDF or PPTX paths into a learning package, assessment, pre-session role-play kit, or post-session scorecard while retaining source-set isolation and exact page or slide excerpts.

How it helps

  • Stops when any requested source path is missing, unreadable, or unsupported instead of rendering a partial set.
  • Grounds messages, model answers, and approved responses through structured SourceReference records.
  • Validates and saves structured JSON plus self-contained interactive HTML for offline review and facilitation.

How the plugin works

Input

Exact approved source set

Absolute PDF or PPTX paths plus a requested output type and name; every supplied file must validate.

Tools

Ingest, ground, and render

  • ingest_document
  • list_documents
  • search_content
  • get_document_page
  • render_output
Expected output

Grounded learning artifact

Schema-valid JSON and self-contained HTML for a learning package, assessment, role-play kit, or scorecard.

A three-step workflow

01 / INGEST

Lock the source set

Provide the exact approved PDF or PPTX paths and verify every document ID, page, and slide.

02 / GROUND

Draft from excerpts

Create learning or practice content while attaching each claim to a matching source passage.

03 / REVIEW

Validate and render

Inspect sourcing, fair balance, schema validity, interactive behavior, and final rendered fidelity.

Sample input

The fictional YAML represents a path-first learning-package request and includes one simulated approved-source excerpt so the grounding relationship can be inspected. The named PDF is not distributed and the YAML is not a runtime source document; production ingestion supports approved PDF and PPTX files.

Download fictional input Inspect sample manifest

Interpreted sample output

The representative learning-module excerpt keeps benefit and safety context connected and cites the exact fictional sentence on page 3. Reviewers should compare that excerpt with the ingested page, then inspect the complete source set, message balance, and rendered artifact before field use.

Download representative output JSON

Business value indicators

Indicator What a business reviewer learns
Supplied versus ingested documents Whether the full approved source set entered the request
Claims with exact source excerpts Whether each message can be traced to a page or slide
Unsupported or inaccurate flags Which statements need revision or removal
Safety and fair-balance coverage Whether review should challenge an unbalanced learning story
Rendered-artifact QA exceptions Whether content is hidden, clipped, stale, or interaction-dependent

Boundaries and human review

Use only the exact approved source documents for the request. Generated learning and role-play materials remain drafts for qualified medical, legal, regulatory, learning, and accessibility review; they are not permission for field use. Review off-label risk, unsupported claims, fair balance, source-set isolation, exact excerpts, and final rendering. Every public-beta output must be validated after content changes.

Get started

Inspect the fictional source-to-message trace, then use the canonical pinned guide for supported file types, output schemas, path-first prompts, and renderer behavior.

Open the pinned Field Training guide →

Utility links