Define your learners. Structure your course. Prepare each video.
Define learners, structure a course and prepare one detailed HeyGen prompt per educational video, with narration, scene timing and optional slide content in ChatGPT or Claude.
Educator Course Director is a careful course-planning toolkit for independent educators, adult trainers, subject-matter experts, coaches and small training teams. It helps turn what you know into an audience-specific course structure, reviewed lesson drafts and detailed instructions for educational videos presented by one HeyGen avatar.
The primary deliverable is one self-contained text prompt per video. Each prompt can bring together the requested spoken script, scene direction, estimated timing and optional teaching-slide content, so the educator can review it and transfer it manually to HeyGen Video Agent.
Version 2.0 is a reusable AI instruction package with an offline planning workspace—not a standalone AI model, hosted course platform or automatic video service. The educator remains in control of the teaching, review and production decisions.
A business, training team or individual may commission a course to address a defined workplace or personal need. The buyer’s goals and constraints help frame the brief, but do not automatically describe the people who will learn.
Lesson choices should reflect learner roles, prior knowledge, barriers, working conditions and the performance expected afterward. When those details are uncertain, the workflow can label hypotheses and ask focused questions; it does not turn assumptions into learner research.
It suits people willing to work with an AI assistant, check subject matter and review every generated video. It is a poor fit when you expect a finished, published course without preparation—or when learning depends on supervised practice, a real demonstration, lab work or actual screen recordings the workflow cannot supply. Accredited, regulated and high-consequence education needs qualified reviewers and applicable standards beyond this toolkit.
The workflow moves from a teaching brief toward an observable course plan. It asks for the topic, expertise, authorized sources, business context and constraints, then distinguishes the commissioning client from the intended learner. It can help specify starting knowledge, problems, desired performance, barriers and learning environment before detailed lesson writing begins.
Course structure is organized around learning outcomes and appropriate evidence of performance. The package can track outcome coverage, source references, estimated workload and recorded human content reviews, helping make unfinished planning more visible.
Alignment and completeness checks are workflow aids. They do not establish factual accuracy, instructional quality, permissions or learner mastery.
Ask for an outline, selected lessons or optional full-course drafts. Larger courses can be divided into named, staged batches, with remaining work made explicit rather than implied to fit unlimited AI context or memory.
When requested, companion drafts may include worked examples, practice tasks, feedback, lesson assessments, learner guides, a glossary, a final performance task, a rubric, follow-up work, educator answer keys and a descriptive transcript draft.
These are drafts for educator review. Assessments and answer keys do not become interactive HeyGen or LMS features automatically. A transcript planned from narration and visuals must be checked against the finished video before it can represent that video accurately.
Every video prompt is intended to stand on its own rather than depend on HeyGen retrieving a separate course document. Depending on the lesson, it can include:
Timing and visual directions are requests, not guarantees. A prompt cannot force exact duration, camera perspective, gestures, emotional performance, typography, captions or slide layout. Check estimated timing against the actual result and adapt to the controls available in your own HeyGen account.
The default planning setup is one avatar, a consistent voice, 16:9 video, calm and encouraging delivery, fixed eye-level medium framing, a quiet neutral background and optional slides. A 1080p preference applies only if supported by the user’s account. The initial English narration estimate is 125 words per minute, plus requested pauses and holds. A three-to-six-minute explanation is one possible planning starting point where appropriate—not a universal lesson-length rule. Slide word-count warnings and illustrative layout guides are adjustable design aids, not official HeyGen limits or safe zones.
The included Prompt Studio is a self-contained HTML form. It contains no AI model and does not create course content on its own: users bring material from their assistant conversation into the form, or open a compatible course JSON record prepared in that conversation.
The workspace keeps the HeyGen prompt separate from narration-only output and creator notes. It also preserves source references, outcome coverage, estimated workload and recorded human review information for the course record.
Browser saving is off by default. If enabled, it uses unencrypted local storage; it is not a backup. Download course records and retain them separately.
The workspace opens its own schema 2.0 course JSON. It does not import HeyGen projects or exports, and version 1.0 course records are not automatically migrated. Keep original records and prepare compatible 2.0 data rather than expecting an automatic conversion.
Recorded human reviews can become stale when relevant content changes. This is local change detection—not an authenticated signature, factual certification, accessibility approval, permission clearance or guarantee of instructional quality.
The default learner export excludes designated educator answer keys and delayed feedback. That exclusion is a useful starting point, not a complete leakage safeguard: inspect narration and all other free-text fields for answers or educator-only information before sharing.
Use authorized sources and minimize or anonymize learner information where appropriate. The workspace may be offline, but the overall workflow is not necessarily private or offline: material pasted into ChatGPT, Claude or HeyGen is handled by those separate providers under their applicable settings and terms.
You need suitable access to an AI assistant and separate HeyGen access for generation. You supply or verify any necessary screenshot, diagram, source, asset and permission. The package does not connect to apps, access a HeyGen account, call APIs, import HeyGen material, generate or download video, inspect or edit renders, publish content, spend credits or manage subscriptions.
It also does not provide a finished slide deck, LMS, SCORM package, interactive player, enrollment system or certification process. It does not replace learner research, subject expertise, qualified review or production checks. ChatGPT, Claude and HeyGen subscriptions, generation credits, avatar and voice rights, and third-party asset rights are not included.
An optional native skill package is included with manual installer instructions where the host supports it. The package also provides a portable route; having an archive does not mean the skill is installed or active.
Native skill saving was attempted automatically and failed. Use the supplied manual installation instructions where supported, or use the portable workflow. Host and account controls may differ.
The six-video example is fictional, not a client case study, testimonial, real HeyGen footage or evidence of educational outcomes. Native installation and activation in an actual ChatGPT or Claude interface are not established by the included package.
The version 2.0 validation record reports 30 local core/command-line checks plus 13 generated-HTML event-handler checks, 43 local checks total. These are package-reported results, not independently verified by MiReVerse. The event-handler checks ran in an in-process DOM harness; they are not a real-browser test.
Two limited assistant trials exercised small fictional educational workflows. They show that instructions and the course-record process could be followed in those particular examples, not that typical users will achieve a given result.
No time savings, revenue increase, engagement improvement or learning gains have been measured or guaranteed. Manual transfer and revision may offset preparation effort. A small pilot is the practical way to assess fit.
There are no integrations, autonomous generation or publishing, bundled subscriptions or generation credits, or measured benefits. No license, support entitlement or future-update commitment is established by this product description. The educator remains responsible for factual accuracy, teaching suitability, permissions, assessments, learner needs and checking every final video.