The concept you can't quite explain back. Built from what you already know.
Thirty-three templates for actually understanding something instead of skimming past it: an explainer built from what you already love, a study plan sized to the weeks you really have, the questions worth asking the one person who already knows this. You give it your real starting point. It builds from there, flags what it can't verify, and never pretends a guess is settled fact.
Works with ChatGPT · Claude · Gemini · whichever AI you already have open
intuition first, jargon last
free to start, no card
If you typed one of these into Google mid-question, this is the page.
- explain this to me the way I'd actually understand itConcept Explainer
- build me a study plan that actually fits my weekTeach Me This Idea Over Several Weeks
- what should I ask the one person who actually knows thisGive Me The Questions To Ask The Person Who Knows This
- map the different camps before I pick a side on thisGet The Shape Of An Idea Before I Dig In
- check whether this claim I read is actually solidPressure-Test This Claim About The Idea
- turn my scattered instincts into a framework I can reuseBuild Me A Way To Think About This Topic
- research this just enough to decide, then stopResearch This Topic For A Real Use, Then Stop
- figure out what I'm actually curious about before I start researchingHelp Me Figure Out What I Am Actually Curious About
Eight of thirty-three. Explaining, pacing a study plan, prepping for the one conversation that matters, and knowing what you actually believe before you build on it.
What you type in about what you already know. What comes back built from it.
Made the way you would make them: your real question, your real week, run once on the AI, not cleaned up after. The name and the details are invented. The reasoning is not.
Why the sourdough came out dense again, explained through home coffee roasting
The one-sentence version
You let dough rest so the gluten network can relax and the yeast-made gases can spread evenly, giving the loaf room to puff up instead of staying dense.
The misconception
Many bakers think “big = ready,” but size alone doesn't guarantee that gluten has relaxed or that bubbles are evenly distributed. A dough that's puffed but still tight will collapse in the oven, giving a dense loaf.
Where it shows up this week
- Your next sourdough batch: Set a timer for the final proof (usually 2-4 h at room temperature or overnight in the fridge). When you feel the dough, note how much softer it is compared to last time.
- Muffin or pancake batter: Even quick breads benefit from a short rest (10-15 min). You'll see fewer dense spots if you let the gluten relax briefly before cooking.
Plus the load-bearing analogy built from coffee roasting (and the one honest sentence on where that analogy breaks), the full mechanism in plain terms (gluten relaxation, bubble growth, acid development), and the explain-it-back prompt to check it actually landed.
Understanding compound interest well enough to plan real retirement contributions, in five weeks
The path (two of five weeks)
| Week | Focus | Checkpoint |
|---|---|---|
| 1 | Grasp the basic “interest-on-principal” idea and the difference between simple vs. compound interest. | Explain in your own words why money left to grow compounds, and compute a simple 1-year compound amount by hand. |
| 4 | Explore “early vs. late” contributions and the impact of compounding frequency (annual, semi-annual, monthly, daily). | Show numerically how contributing the same total amount earlier yields a larger final balance than later; compare monthly vs. annual compounding. |
This week's focus
- Read a short article or watch a 5-minute video on “simple vs. compound interest”, focus on the intuition: “interest earns interest.”
- Do the hand calculation: If you invest $1,000 at 5 % annual simple interest for 3 years → $1,150. Then compute the same with compound interest (annual compounding) → $1,157.63.
If a week gets missed
Skip Week 2: merge Weeks 1 & 2 by using a combined “compound-interest formula” cheat sheet; focus only on plugging numbers rather than deriving the algebra.
Plus the full five-week path through regular contributions and the personal retirement sketch in week five, and a checkpoint at every stage so you know the understanding is actually taking hold, not just the calendar moving forward.
Fifteen minutes at Sunday dinner to ask an electrician uncle about a DIY panel upgrade
The ranked questions
- "Do local codes allow a homeowner to pull the permit and perform the panel work themselves if they're not a licensed electrician?" Reveals legal permissibility vs. professional-only rule.
- "When an inspector signs off on a DIY panel upgrade, what specific documentation or evidence must I provide to prove compliance?" Shows practical hurdles after the permit.
The opening question
Uncle, in our area can I actually pull the permit and do the panel upgrade myself, or must a licensed electrician handle it?
What a weak answer would reveal
- Weak answer ("I think you can, but check the city website"): Uncle isn't sure of the rule and may be avoiding liability.
- Weak answer ("Just bring the permit"): He's unaware of required inspection reports or as-built diagrams.
Plus the short version trimmed to fit fifteen minutes, the follow-up questions if the opening one lands well, and what a weak answer on liability would reveal about the rule itself.
Three steps. No blank page.
Name what you're actually trying to get
Explain a concept, pace a study plan across real weeks, or build the question list for the one conversation that matters.
Give it what you already know
Four to seven blanks: the topic, your current level, what sparked the question, how deep to go. It builds from there, not from zero.
Send it to your AI, then use it
Ultra Prompt builds the expert prompt and hands it to ChatGPT, Claude, or Gemini. Anything it can't verify from what you gave it gets flagged, never guessed.
Templates explain one thing. An agent can keep teaching you.
The same facts you type into a template can live inside an AI agent that knows what you already understand, how you actually learn, and the question you're circling, so every explanation and every study session starts from where you really are instead of from scratch. Three agents worth building:
The Concept Bridge
Knows what you already love and what you've already been taught, so every new concept gets built from an analogy that actually fits you, with the misconception named before you can pick it up.
Runs in Claude Projects, a Custom GPT, or a local modelThe Pacer
Holds your real weeks and hours, not an idealized study schedule. Tell it a week got missed and it re-sequences the plan instead of making you start the syllabus over.
Runs in your phone, through the AI you already useThe Question Coach
Remembers what you already asked the expert last time, so the next conversation moves past ground you've already covered instead of repeating it.
Runs in Gemini, ChatGPT, or ClaudeBuild your own tutor in an afternoon.
Agent Studio is in every Pro seat. Give it what you already know, how you actually learn best, and the thing you're trying to understand. Then bring it a new question or the next study session; back comes the explanation or the plan, pitched to where you actually are, with anything it can't verify marked plainly. Export it as a Claude Project, a Custom GPT, a Gemini Gem, Copilot, or a local model on your own laptop. It runs on the AI you already have. Nothing to host.
Open Agent StudioOr have it built around what you're learning.
Want it built for you? A custom build is one conversation about what you do every week, then an agent brief written for you, exported to your AI, with a walkthrough. From $450, one time.
Ask about a custom buildFiles are yours to keep. The library, updates, and support ride the subscription.
One plan. Honest about what it does.
Unlimited generates, all 33 learning templates and the full 1,000+ library, every technique and pipeline, and Agent Studio.
Start with 10 free promptsFree to start: 10 finished prompts a month, no card. Cancel anytime.
Build a tutor that knows what you already understand and how you actually learn, then export it to Claude Projects, a Custom GPT, Gemini Gems, Copilot, or a local model. Your own understanding, on your own account.
See Agent StudioThe questions you ask before you trust a tool to explain something.
Do I need ChatGPT or Claude to use this?
Yes, one of them, and the free tier of any of them works. Ultra Prompt builds the prompt and sends you there with it. You never paste anything by hand unless you want to.
Will it just make up facts to fill in the gaps?
It's built not to. The templates tell the model to mark anything it can't verify from what you typed instead of inventing a statistic, a study, or a source to sound complete.
How is this different from just asking ChatGPT to explain something?
You can ask ChatGPT to explain a concept and get the textbook version. These templates ask what you already love, what sparked the question, and how deep to go, then build the explanation from your own anchor instead of a generic one.
Will the study plan actually fit my real schedule?
It's built from the hours you say you actually have, not an idealized week. Miss a week and the same template tells you what to cut or combine, not what to feel bad about.
Is this a substitute for a real teacher or expert?
No. It explains, paces, and preps you for the conversation. The question-prep templates exist specifically to make the time with a real person count for more, not to replace them.
What happens to what I typed in?
Nothing you typed is sent anywhere but your own AI chat. Save the output if you want a record; Ultra Prompt doesn't keep a copy of what you were trying to learn.