You've done the needs analysis. You've got the stakeholder buy-in. You know exactly what the program needs to achieve, which Bloom's levels the learning objectives need to hit, which modalities make sense for the audience, and what success metrics you'll report back to the business. And you're still staring at a blank facilitator guide, a half-done course outline, a stack of SME notes that need to be turned into actual learning content, and a learning impact report due Friday.
That gap — between knowing what a program needs to be and getting the words on the page — is where most of your time goes. Not the strategy. Not the design thinking. The documentation layer: the prose that wraps every design decision you've already made. Claude is a large language model that's excellent at closing that gap. You give it structured notes, content bullets, interview transcripts, or data — and it turns that into polished instructional writing. It is not an LMS, not an authoring tool, not a video production platform. It is a drafting engine you drive.
Before the use cases, the caveat you need: Claude cannot access your LMS — not Cornerstone, not Workday Learning, not SAP SuccessFactors, Docebo, or TalentLMS. It cannot connect to Articulate, Lectora, or Rise. It cannot query your assessment data in Qualtrics or SurveyMonkey, access your HRIS, read learner records, or pull completion tracking. It cannot build SCORM packages, generate video, record narration, or publish anything anywhere. It is strictly a drafting and writing tool. You bring the expertise, the SME knowledge, the learning strategy — Claude handles the prose. Every prompt below starts with "paste in your content" for exactly that reason.
1. Course Outline and Learning Objectives
The first document every program needs — and the one that takes longest to produce from raw SME input. You've got the content domain covered. You know the audience, the performance gap, and roughly how many modules make sense. What you need is a structured outline that maps content to Bloom's levels, assigns durations, and surfaces where application activities are missing before you get deep into development.
Paste your SME notes or content bullets. Claude produces the structured outline.
I'm building a course outline for [topic]. Target audience: [describe — role, experience level, what they'll do with this on the job]. Performance goal: [what they need to be able to do after completing this program]. Here is the raw content: [paste your SME notes, content bullets, or source material].
Build a structured course outline including:
- Module titles with Lesson titles under each module
- Learning objectives for each lesson written to Bloom's taxonomy levels — identify the appropriate level (remembering, understanding, applying, analyzing, evaluating, creating) and label each objective accordingly
- Estimated duration per module (ILT, eLearning, and/or blended)
- Recommended modality per module: ILT, eLearning, practice activity, job aid, or combination — with a one-sentence rationale for each
- Flag any modules where application activities are missing and the objectives require applying-level or above
Format as a structured table where useful. Do not invent content not present in the notes I provided.
The flag on missing application activities is the signal you want early — before the stakeholder review, not after development is complete. Claude won't fabricate content you didn't give it; if the SME notes have a gap, it'll flag the gap rather than fill it with invented material. For HR Business Partners who co-own capability development investments, Claude AI for HR Business Partners covers the business case and stakeholder communication layer.
2. Facilitator Guide Narrative
Slide decks and outlines tell facilitators what to show. Facilitator guides tell them what to say, how to read the room, what questions to ask, and what misconceptions will surface. Writing that narrative layer — the part that transforms a deck into a guide a facilitator can actually use — is time-consuming and often the last thing that gets done before delivery.
Paste your slide content or session outline. Claude drafts the facilitator guide section.
I'm writing a facilitator guide section for [session name / topic]. Here's the content: [paste your slide content, outline, or session notes — include learning objectives for this section if you have them].
Audience for the guide: facilitators who know the subject but may not have delivered this program before.
Write a complete facilitator guide section including:
- Facilitator Notes: what to say for each slide or section — not a transcript, but key talking points, transitions, and context the facilitator needs that isn't visible in the slide content
- Discussion Questions: 3–5 questions that drive application, not recall — include a facilitation tip for each question on how to draw out quieter participants or redirect a conversation that's going off-track
- Timing Guide: estimated time per activity or section with total session time
- Common Misconceptions & How to Address Them: 3–4 misconceptions learners typically bring to this topic, with suggested facilitator responses
- Debrief Questions: 2–3 questions to close the session and connect learning to on-the-job application
Do not fabricate domain-specific content not present in the materials I provided.
Critical caveat: Facilitator guide accuracy depends on what you give Claude. It will produce the right structure and instructional language — the facilitation tips, the debrief framing, the misconception handling — but the domain-specific content must come from your materials. Review every talking point against the actual content before handing this to a facilitator.
3. eLearning Script
Narration scripts are one of the highest-effort, lowest-excitement artifacts in instructional design. The content is clear, the learning objectives are written, the visual treatment is decided — and you still have to produce 800–1,200 words of narration in the right tone, in split-screen format, with a scenario setup and a defensible knowledge check.
Paste your content outline, objectives, and tone guidance. Claude produces the draft script.
I'm writing an eLearning narration script for [module name]. Learning objectives for this module: [paste objectives]. Tone: [conversational / formal / scenario-driven — describe the voice]. Target audience: [role, experience level, where they'll apply this learning]. Here's the content outline: [paste your content bullets or outline].
Write a narration script in split-screen format with two columns: Visual Description | Narration Text.
Include:
- Intro hook (first 30 seconds — a scenario, question, or situation that makes the learner care about this content before you tell them what they're about to learn)
- Scenario setup: a realistic workplace scenario that the learner will navigate through the module
- Knowledge check question at the midpoint or end: write the question, 4 answer options, correct answer with explanation, and distractor analysis (why each wrong answer is plausible and what misconception it addresses)
- Closing summary: connect back to the learning objectives and the opening scenario
Do not invent technical content, data, or examples not grounded in the materials I provided.
Critical caveat: Scenario accuracy is everything in eLearning. Claude can build the instructional scaffolding — the hook, the scenario arc, the knowledge check structure — but the scenario details must reflect how the work actually happens. Validate every scenario element with your SME before sending to development.
4. Learning Needs Analysis Summary and Stakeholder Report
You've done the interviews, reviewed the performance data, identified the gap, and formed a view on whether L&D is even the right intervention. Now you need to turn that into an executive summary that a business stakeholder — who hasn't seen any of your process — can read in five minutes and use to decide whether to fund the program.
Paste your interview notes and performance gap data. Claude produces the executive summary.
I've completed a learning needs analysis for [department / role / performance area]. Here are my interview notes and performance gap data: [paste your interview notes, survey data, performance metrics, manager feedback — whatever you collected].
Write a needs analysis executive summary for a business stakeholder audience (not L&D — people who are deciding whether to fund this program). Include:
- Performance Gap: what the data shows — specific and measurable where possible
- Root Cause Analysis: categorize the cause — knowledge/skill gap, motivation gap, environmental/systems barrier, or combination — with evidence from the data for each
- Recommended Intervention: L&D solution recommendation with rationale, OR a recommendation that this is not a training problem (and what the actual intervention should be) — with the reasoning behind that call
- Proposed Success Metrics: how you'll measure whether the intervention worked — leading and lagging indicators
- Estimated Scope: rough program size, modality, audience count, and timeline
Tone: direct, evidence-based, business-facing. Not jargon-heavy L&D language — write for a business stakeholder who will ask "why does this need to be training?"
The "not a training problem" section matters. Claude will include it if your data points that direction — which is exactly the conversation you want to have with the business before you build something that won't move the metric. For HR and People Operations teams who initiate many of these requests, Claude AI for HR and People Operations covers the HR side of that workflow.
5. Post-Training Evaluation and Learning Impact Report
The evaluation data is in. Completion rates are up. Survey scores look reasonable. Manager feedback is mixed. Now you need to turn that into a report the CLO or business stakeholder can actually read — not a data dump, but a narrative that tells them what worked, what didn't, and what you're doing about it.
Paste your survey data, completion stats, and manager feedback. Claude produces the impact narrative.
I'm writing a post-training evaluation report for [program name]. Here's the data: [paste your survey results — aggregate scores and open-text themes; completion statistics; manager observation feedback or 90-day follow-up data if you have it].
Write a learning impact report for a [HR leadership / CLO / business stakeholder] audience. Include:
- What We Measured: evaluation methodology — Levels 1–3 or 1–4 if you have the data, with what was collected and how
- Key Findings: what improved (with data), what didn't (with data), and the honest interpretation
- Learner Feedback Themes: top 3–4 themes from open-text survey responses — positive and constructive
- Manager Observation Summary: what managers reported about on-the-job behavior change (or lack of it)
- Recommended Adjustments for Next Cohort: specific program changes based on the findings
- ROI Framing: if the data supports a business impact connection, frame it — if it doesn't, say so rather than overclaiming
Tone: honest and evidence-based. Do not inflate findings. If the data is inconclusive, say it's inconclusive.
Critical caveat: Paste de-identified, aggregate data only — no individual learner names, employee IDs, or any data that could identify a specific person. Claude drafts the narrative from what you provide; your review of the framing before distribution is essential. For finance and operations stakeholders reading the ROI section, Claude AI for Business Analysts covers the data narrative workflow on the receiving end.
6. SME Interview-to-Content Conversion
This is the one that saves the most time — 3–4 hours per SME session, which is the most time-consuming single step in most instructional design workflows. You've done the SME interview. You have a transcript or detailed notes. The knowledge is in there. What you need is someone to extract it, sequence it, flag what's missing, and turn it into content you can actually build from.
Paste your SME interview transcript or session notes. Claude does the extraction.
I've completed an SME interview for [topic / course module]. Here is the transcript or notes: [paste your full SME interview transcript or detailed notes].
From this material, produce:
- Key Concepts Extracted: a numbered list of the core concepts the SME covered, in plain language
- Proposed Logical Sequence: reorder the concepts into a learning sequence that moves from foundational to applied — with a one-sentence rationale for the sequence
- Knowledge Gaps Flagged: concepts that appear in the interview but are underdeveloped, contradictory, or missing the "so what for the learner" — note what you'd need to go back to the SME to clarify
- Suggested Examples and Analogies: 3–5 examples or analogies the SME mentioned (or that are implied by what they said) that would make the concepts land for learners
- Draft FAQ for Learners: 5–7 questions a learner would realistically ask after completing this module, with draft answers based on what the SME covered
Do not invent content not present in the transcript. Flag gaps rather than fill them.
The "flag gaps rather than fill them" instruction is critical — it's what makes this actually useful. You want a list of questions to take back to the SME, not invented content that sounds right but isn't. The output here isn't the finished content; it's the structured draft that makes your next pass 3–4 hours faster.
Why Claude Over ChatGPT for L&D Work
100K+ context window. Paste an entire SME interview transcript — 10,000 words — and Claude processes the full thing. Paste a complete course worth of content, a full quarter of learner feedback, or a whole needs analysis package. No truncation, no "please summarize first." The full context is in the room for every output you produce.
Conservative where it matters. Claude won't fabricate research citations to support your needs analysis framing. It won't assign Bloom's taxonomy levels it's not confident about. It won't invent compliance language or regulatory requirements you didn't provide. In L&D work — where accuracy of content and defensibility of design decisions matter — that conservatism is a feature, not a limitation.
Projects per learning program. Build one Claude Project per program or curriculum. Load the needs analysis, the audience profile, the learning objectives, and the SME notes into the project system prompt. Now every artifact you produce — outline, facilitator guide, eLearning script, evaluation report — is produced with the full program context in scope. The project knows the learning objectives when you're writing the post-training evaluation. That coherence matters.
Structured output fidelity. Split-screen eLearning scripts render correctly every time. Facilitator guide tables come back in the right format. Module-by-module course outlines with Bloom's labels and duration estimates don't collapse into prose. Claude holds complex document structures across long outputs — which is what L&D documentation actually requires. For a direct comparison on this and other capabilities, Claude vs ChatGPT: Which AI Is Better for Work covers the tradeoffs.
Practical Tips for L&D Professionals
One Claude Project per learning program. Load the needs analysis, SME notes, audience profile, and learning objectives into the project system prompt before you start building. Every artifact you produce from that point — outline, facilitator guide, eLearning script, knowledge checks, evaluation instruments — is produced against the same goals. No re-explaining the context. No drift between documents.
SME expertise in / structure out. This is the most important mental model. Your job is to bring the knowledge and validate every accuracy claim; Claude's job is to turn that knowledge into instructional prose. Never let Claude invent content it wasn't given. The "do not fabricate content not present in the materials" instruction in every prompt above exists for this reason — enforce it explicitly every time.
Specify the learner and the performance context. "Write for a new hire with 0–3 months experience who will use this on the warehouse floor, not in a classroom" produces dramatically better content than "write a training script." The audience specificity shapes the vocabulary, the examples, the scaffolding level, and the length of narration sentences. Give Claude the learner profile and the performance context every time.
End-of-program documentation sprint. After delivery, block two hours. Paste your debrief notes, facilitator feedback, and evaluation data into Claude and produce the impact report, lessons learned doc, and program recommendations in one session. Not two weeks later when the context has faded — the day after the cohort closes, while the details are fresh. It turns a half-day of writing into 45 minutes of editing. For executive assistants managing the calendar and reporting workflows around these sprints, Claude AI for Executive Assistants covers the adjacent documentation workflow. For operations managers who receive the program outputs and workforce capability reports, Claude AI for Operations Managers is the parallel read.
The Playbook
If you want every prompt above — refined, tested, and organized by L&D workflow — plus prompts for program communications, stakeholder presentations, SME prep guides, change management messaging, and the full library of professional use cases, that's what The Complete Claude Playbook is.
$27. Instant access. The prompts your L&D work actually needs.