AI for Learning

Chapter 01 · Getting started

A practical field guide to learning with AI.

Six strategies for putting AI to work in how you learn, teach, and build — each with real examples and prompts you can copy. We default to Claude, but the ideas carry to any capable assistant.

Pick a strategy

Jump straight to the area you’re working in — content, feedback, design, code, analytics, or analysis. Each chapter stands on its own.

Copy the prompts

Every chapter includes a ready-to-use prompt. Paste it in, swap in your own details, and you have a running start instead of a blank page.

Iterate, play & experiment

Treat the first answer as a draft. The real value comes from the back-and-forth — refine, push back, and ask for another pass.

Default to Claude

We reach for Claude first. It works across every one of our teams — UX, design, development, feedback, and analysis — and we find it especially strong at long-form reasoning, careful writing, and brainstorming alongside the people doing the work. ChatGPT and Gemini are great too.

Try Claude

Chapter 02 · Introduction

AI is the most flexible learning tool you’ve ever had.

It drafts, critiques, explains, designs, codes, and analyzes — on demand, at whatever level you need. Used well, it doesn’t replace your thinking; it gives you a tireless partner to think out loud with, shortening the loop between an idea and something real.

From consumer to creator

The same tool that answers a question can help you build the lesson, the prototype, or the report — moving you from reading about a thing to making one.

A patient thinking partner

Ask the “dumb” question. Ask for it five different ways. Ask it to explain like you’re new, then like you’re an expert. It never tires of the follow-up.

Faster loops

Outline, draft, critique, revise — steps that used to take days collapse into a conversation. You spend your time judging and shaping, not staring at a blank page.

Claude Try this to start a learning session

I want to learn [topic] well enough to [goal] in [timeframe].
I currently know [your level / background].

Act as a patient tutor. Start by asking me 3 questions to
gauge where I am, then propose a short learning plan with
checkpoints. Keep explanations concrete and use examples.

Why it works: you give Claude your goal, level, and constraints, then ask it to diagnose before it teaches — so the plan fits you instead of a generic outline.

Try

Talking to your AI

One of the most fun, helpful, and amazing ways to collaborate is simply to talk to it. Switch on voice mode and think out loud while you walk, work, or sketch — ask questions, brainstorm, and work an idea back and forth like you would with a colleague. Available in Claude, ChatGPT, and Gemini.

Strategies for using AI

Six ways to put AI to work.

From first draft to final analysis, here’s where AI earns its place in a learning workflow — with a concrete example and a copy-ready Claude prompt for each.

Strategy 01

Content Development

Draft, structure, and shape learning content — from a blank page to a working module.

View Strategy 02

Feedback

Specific, actionable critique on the work — scored against the bar you set.

View Strategy 03

Design

Design the experience, not just the content — structure, storyboards, accessibility.

View Strategy 04

Code

Build, debug, and learn to code — with a tutor and a prototyper on call.

View Strategy 05

Analytics

Make sense of your learning data — summaries, queries, trends, and cohorts.

View Strategy 06

Analysis

Go deeper — synthesize sources, weigh trade-offs, and turn it into a recommendation.

View

Strategy 01 · Content development

Draft, structure, and shape learning content.

The blank page is the slowest part of making anything. AI gets you to a working draft fast — an outline, an explainer, a script — so your effort goes into refining rather than starting.

Outlines & lesson plans

Turn a topic and an audience into a structured module — objectives, sections, activities, and a knowledge check — in one pass.

Explainers at the right level

Ask for the same concept for a beginner, a manager, and an expert. Pick the one that lands, or splice the best of each.

Repurpose across formats

One source — a doc, a transcript, a deck — becomes a summary, a quiz, an email, and a set of slides without rewriting from scratch.

Here are my rough notes on [topic]:
[paste notes]
Turn this into a 20-minute lesson for [audience]. Give me:
1. Three learning objectives
2. A section-by-section outline with timing
3. One hands-on activity
4. Five quiz questions with answers
Flag anything in my notes that's thin or needs a source.

Why Claude: Reasons over messy notes and can build the quiz as an Artifact.

I'm attaching my notes/slides on [topic] (file or pasted below).
[paste or upload]
First, read them and pull out the key points. Then build a 20-minute lesson for [audience]: 3 objectives, a timed outline, one activity, and 5 quiz questions with answers.
If a stat or claim needs checking, browse the web and cite a current source.

Why ChatGPT: Upload the file, then let browsing verify claims and refresh sources.

Open my Google Doc "[doc name]" with my notes on [topic].
Restructure it in place into a 20-minute lesson for [audience]: 3 objectives, a timed section outline, one activity, and 5 quiz questions with answers.
Keep my original notes below a divider so I can compare.

Why Gemini: Acts directly inside your Google Doc and handles long source material.

Using my Word document [file name] with notes on [topic], draft a 20-minute lesson for [audience].
Include 3 objectives, a timed outline, one activity, and 5 quiz questions with answers, formatted with headings I can reuse.
Then offer to turn the outline into a PowerPoint starter deck.

Why Microsoft Copilot: Works on your real Word file and spins the outline into PowerPoint.

Strategy 02 · Feedback

Get specific, actionable feedback on the work.

AI is an honest, always-available reviewer. Give it your draft and the bar you’re aiming for, and it will tell you what’s working, what’s missing, and what to fix — without the wait or the ego.

Critique against a rubric

Paste your criteria and the draft. Get a line-by-line review scored against the bar you actually care about.

Spot what’s missing

Ask what a skeptical reader would still wonder, where the logic jumps, or which claim needs evidence. It finds the gaps you’re too close to see.

Pressure-test your thinking

Have it argue the other side, or play a tough stakeholder. Better to meet the hard questions here than in the room.

You're a demanding but constructive reviewer.
Here's what "great" looks like for this piece:
[criteria / rubric]
Here's my draft:
[paste draft]
Score it against each criterion (1-5), quote the exact lines that fall short, and give me the single highest-impact fix first.

Why Claude: Rigorous close reading and exact line-level critique against a rubric.

You're a demanding but constructive reviewer. I've uploaded my draft and a rubric file.
[upload draft + rubric]
Score each criterion 1-5, quote the lines that fall short, and lead with the highest-impact fix.
Then output the scores as a table and a simple bar chart I can download.

Why ChatGPT: Reads uploaded files and visualizes the scores as a chart.

Review my Google Doc "[doc name]" against this bar:
[criteria / rubric]
Score each criterion 1-5, and add a comment on the exact lines that fall short using Doc comments.
Give me the single highest-impact fix first in the chat.

Why Gemini: Drops feedback as native comments right on your Doc's weak lines.

In this Word document, review my draft against the bar below:
[criteria / rubric]
Score each criterion 1-5, quote the lines that fall short, and suggest tracked-change edits for the highest-impact fix first.
Summarize the scores in a table at the top.

Why Microsoft Copilot: Reviews in Word and proposes the fixes as tracked changes.

Strategy 03 · Design

Design the experience, not just the content.

Good learning is designed — how it’s sequenced, how it looks, who it leaves out. AI is a fast collaborator for instructional structure, visual direction, and accessibility checks before you commit.

Instructional design

Sequence a course or module, map content to outcomes, and balance instruction with practice and retrieval.

Storyboards & wireframes

Describe a screen or activity and get a layout, copy, and interaction notes you can hand to a designer or prototype yourself.

Accessibility & inclusion

Check reading level, contrast, alt-text, and inclusive language — so the experience works for more of your learners by default.

I'm designing a self-paced module on [topic] for [audience].
Constraints: [time limit, device, tone].
Propose a storyboard: for each screen give a title, the on-screen content, the learner action, and an accessibility note. Aim for active learning, not just reading. Then suggest one interaction that would make it more memorable.

Why Claude: Thinks through learning design and can prototype the interaction as an Artifact.

I'm designing a self-paced module on [topic] for [audience]. Constraints: [time, device, tone].
Build a storyboard: per screen give a title, on-screen content, learner action, and an accessibility note, output as a table.
For the key concept screen, generate a clear instructional image with DALL-E I could use as the visual.

Why ChatGPT: Tables the storyboard and generates a sample visual with DALL-E.

I'm designing a self-paced module on [topic] for [audience]. Constraints: [time, device, tone].
Build the storyboard as a Google Sheet: one row per screen with columns for title, on-screen content, learner action, and accessibility note.
I'll also paste a screenshot of a reference screen, tell me what to keep or fix for accessibility.

Why Gemini: Builds the storyboard in Sheets and reads your reference screenshots.

I'm designing a self-paced module on [topic] for [audience]. Constraints: [time, device, tone].
Draft the storyboard in this Word doc: per screen list title, on-screen content, learner action, and accessibility note.
Then create a PowerPoint where each slide is one screen, and check it with the Accessibility Checker.

Why Microsoft Copilot: Drafts in Word, builds the slides, and runs PowerPoint's accessibility check.

Strategy 04 · Code

Build, debug, and learn to code — with a tutor on call.

You don’t have to be an engineer to make working things. AI can prototype an interactive demo, explain an error in plain language, or walk you through a new language one step at a time.

Prototype interactive demos

Describe an idea and get a runnable HTML/JS widget — a quiz, a calculator, a simulation — to test before you invest.

Debug & explain errors

Paste the code and the error. Get a likely cause, a fix, and — just as useful — an explanation of why it broke.

Learn a language or framework

Ask for a guided path with small exercises, then have it review your attempts and nudge you toward better habits.

Build a single-file HTML page (inline CSS + JS, no libraries) that's a [flashcard / quiz / calculator] for [topic].
Requirements: [list them]. Keep it clean and mobile-friendly.
After the code, explain how it works in plain language and suggest two ways I could extend it.

Why Claude: Artifacts render the interactive demo live and explain the code clearly.

I'm learning [language/framework] and want to build [small project, e.g. a quiz app].
Write the code, then run it to show the output and catch errors before I paste it.
If I upload my [file/CSV], use it as the real data source.
Walk me through each function, then give me one debugging exercise.

Why ChatGPT: Code Interpreter actually runs the code and tests it against your files.

Here's a screenshot of my [code / error / UI bug]: [attach image].
Tell me what's wrong and the exact fix. Then give me a clean version I can paste into Google Colab to run free in the browser.
Add inline comments so a beginner can follow each step.

Why Gemini: Reads screenshots of your code and bugs, then hands off to free Colab.

In VS Code, look at the open file [filename] in my repo.
Explain what this function does in plain language, then suggest a fix for [bug/behavior] and write a unit test for it.
Use the existing patterns and naming in this repo. Show the diff before applying.

Why Microsoft Copilot: GitHub Copilot reads your actual repo files and edits them in place.

Strategy 05 · Analytics

Make sense of your learning data.

Survey results, quiz scores, completion rates, engagement logs — the signal is in there. AI helps you query it, summarize it, and turn rows of numbers into something you can act on.

Summarize results

Paste survey or quiz exports and get the headline takeaways, the outliers, and the themes in open-text responses.

Write the query or formula

Describe what you want to know in plain English; get the SQL, the spreadsheet formula, or the pivot setup to find it.

Spot trends & cohorts

Ask where drop-off happens, which group struggles, or what changed after a redesign — and what to look at next.

Here's data from [course / program]: [paste CSV or describe].
Tell me: (1) the 3 most important takeaways, (2) where learners drop off or struggle, (3) any group that's an outlier, and (4) the two changes you'd test next. Show your reasoning, and flag where the data is too thin to be confident.

Why Claude: Careful reasoning over messy data with honest confidence flags.

I'm uploading [quiz_results.csv / completion_data.csv].
Use Advanced Data Analysis to clean it, then chart completion by [cohort] and flag the questions with the lowest pass rates.
Build a correlation table between [engagement metric] and [final score].
Show the charts and give me the Python so I can rerun it next month.

Why ChatGPT: Runs real Python on your uploaded CSV and generates the charts.

This is my learning data in Google Sheets: [share sheet or paste range].
Inside Sheets, suggest the formulas to calculate completion rate, average score by cohort, and week-over-week trend.
Then build a pivot summary and tell me the two cohorts that need attention and why.

Why Gemini: Works inside Google Sheets to write formulas and pivots on your data.

In this Excel workbook, look at the [survey/quiz results] on sheet [name].
Summarize the key trends, add a PivotTable by [cohort], and flag any outlier groups.
Then draft an Outlook email to my team with the 3 findings and a recommended next step.

Why Microsoft Copilot: Acts on your live Excel workbook and drafts the follow-up in Outlook.

Strategy 06 · Analysis

Go deeper — synthesize, compare, and decide.

Where analytics tells you what, analysis works out so what. AI is strong at holding many sources at once — pulling out themes, weighing trade-offs, and turning a pile of material into a recommendation.

Synthesize many sources

Feed it interviews, articles, or reports and get the common threads, the disagreements, and what they add up to.

Compare options & trade-offs

Lay two or three choices side by side against the criteria that matter, with the honest downsides of each.

Turn analysis into a call

Move from “here’s what we found” to “here’s what I’d do and why” — with the assumptions made explicit.

I'm analyzing [question]. Here are the sources:
[paste docs / notes / data]
First, pull out the 4-6 themes that cut across them, with a supporting quote for each. Note where sources disagree. Then give me a recommendation, the trade-offs, and the assumptions it depends on. Tell me what would change your mind.

Why Claude: Long-context synthesis that weighs trade-offs and surfaces its assumptions.

Research [question] for me. Browse the web for [3-4 recent sources/vendors] and cite each with links.
Compare them in a table by [criteria: cost, fit, evidence].
Then give a recommendation, the trade-offs, and what you'd verify before deciding.

Why ChatGPT: Browses live sources and builds a cited comparison table.

I've put my source documents in Google Drive: [link the Docs/folder].
Pull the 4-6 cross-cutting themes with a supporting quote from each doc, and note where they conflict.
Then draft a one-page recommendation in a new Google Doc with trade-offs and open questions.

Why Gemini: Reads your Drive documents and writes the recommendation into Docs.

Using these files in my Microsoft 365 [Word docs / Teams notes / PowerPoint]: [reference them].
Synthesize the recurring themes and where they disagree, then build a recommendation.
Draft a PowerPoint deck with one slide per theme plus a final recommendation slide I can edit.

Why Microsoft Copilot: Synthesizes across your M365 files and builds an editable deck.

What’s next

From assistant to agent.

So far we’ve used AI as a partner you talk to. The next step is letting it act — agents that connect into your pipeline (your docs, tools, and data), automate the busywork, and deliver help at the exact moment of need, right in the flow of work.

Example 01

Performance support in the flow of work

An agent wired into your knowledge base and tools lives in Slack or Teams. When someone hits a wall, it answers with the exact step, pulls the right job aid, and takes the action — opens the ticket, books the refresher, updates the record — without anyone leaving the conversation.

Connect KB + tools Catch the question Answer + act

Example 02

A self-updating content pipeline

An agent watches where your product changes — GitHub, Jira, Drive. When something ships, it drafts the updated lesson or job aid, routes it to a human for a quick review, and publishes once approved. Your training stays current on its own.

Watch for changes Draft the update Review + publish

Getting started

A great first agentic task: let AI triage your inbox. Connect your email, then have it summarize what came in, pull out your to-dos, and flag the follow-ups you owe — every morning. Keep a human approving anything it sends until you trust it.

Connect Gmail, then:

Scan my inbox from the last 24 hours. Summarize each thread
in one line, list what needs a reply, and draft a short
response for the top three.

Why Claude: careful summaries and ready-to-send drafts in one pass.

Right inside Gmail:

Summarize my unread emails, group them by project, and give
me a to-do list of follow-ups with who I'm waiting on.

Why Gemini: it already lives in Gmail and Workspace — no setup.

With the Gmail connector:

Every weekday at 8am, triage my inbox — what's urgent, what
can wait, and the five follow-ups I owe people.

Why ChatGPT: schedule it as a recurring task that runs each morning.

Apps to try

Start with these.

A short, hand-picked set of AI tools worth your time — for research, creation, and vibe coding. The yellow ★ marks a Nick’s Pick. More in the full AI list on CoolAppsforLearning.com.

★ NotebookLM

Research notebook

Google’s research notebook — upload your sources and get grounded summaries, Q&A, and even an audio overview. Perfect for studying a stack of documents.

Try NotebookLM ↗

★ Claude

Thinking & creation

Our default throughout this notebook. Strong at long-form reasoning, careful writing, and building interactive artifacts you can design and ship right in the chat.

Try Claude ↗

★ Claude Code

Vibe coding

Anthropic’s agentic coding tool that lives in your terminal — plan, build, and debug whole features in plain language while Claude handles the code.

Try Claude Code ↗

★ Lovable

Vibe coding

Describe an app in plain English and get a working full-stack build with live previews — ideal for MVPs, prototypes, and internal tools.

Try Lovable ↗

Cursor

Vibe coding

An AI-first code editor built on VS Code — inline generation, autocomplete, and a chat that understands your whole project. A natural next step from no-code builders.

Try Cursor ↗

Gemini

Multimodal AI

Google’s multimodal assistant — strong across text, images, audio, and code, and tightly woven into Workspace if you live in Docs, Sheets, and Gmail.

Try Gemini ↗

ChatGPT

Versatile assistant

The versatile all-rounder — brainstorm, draft, analyze, and build custom GPTs. A great second opinion to compare against Claude on the same prompt.

Try ChatGPT ↗

Perplexity

AI answers + sources

An answer engine that cites its sources — the one to reach for when you need quick, trustworthy research with links you can actually check.

Try Perplexity ↗

Gamma

AI decks & docs

Turn a prompt into a polished deck, doc, or one-page site in minutes — a fast way to package what you’ve learned into something shareable.

Try Gamma ↗

Additional notebooks

More visual notebooks from Nick Floro and Sealworks — curated, hand-picked, and refreshed regularly to help you design, build, and ship.

mobileappsforlearning.com

Mobile Apps for Learning

create. explore. experiment. share.

Mobile Apps for Learning

A visual notebook of the mobile tools that change how you learn and work — AI assistants, note-taking, collaboration, meetings, annotation, and content creation.

Visit ↗

coolappsforlearning.com

Cool Apps for Learning

apps, tools & resources to level up.

Cool Apps for Learning

The flagship list — 180+ tools across 22 categories that Nick and the Sealworks team reach for every day. Updated monthly.

Visit ↗

prototypingforlearning.com

Prototyping for Learning

sketch. test. evolve. repeat.

Prototyping for Learning

A hands-on field guide to prototyping — why it matters, a simple repeatable process, and the tools we use to test ideas before a single line of code gets written.

Visit ↗

Nick Floro, Co-founder & Chief Learning Architect at Sealworks

Got an idea?Let’s talk.

Whether you’re exploring AI for your first course or rolling it out across a platform, I’d love to hear about it. Coffee chats welcome.

Nick Floro

Sealworks Interactive Studios

Co-founder & Chief Learning Architect. 34+ years turning ideas into prototypes, courses, apps, and platforms — from startups to Fortune 500s, reaching millions of learners every year.

LinkedIn Sealworks