Skip to content
For the decision in front of you right now

The decision you keep circling. Reasoned through, not around.

Sixty-six templates for thinking something through instead of just feeling your way through it: a structured decision framework, a pre-mortem that finds the failure before it happens, a neutral referee for the argument you keep having, a devil's advocate for the belief you're not sure you should trust. You give it the real situation. It reasons from what you gave it, flags what it can't verify, and never pretends a guess is a fact.

Works with ChatGPT  ·  Claude  ·  Gemini  ·  whichever AI you already have open

66thinking & deciding templates
4minfrom blank page to reasoned-through
8failure scenarios in one pre-mortem
each with a warning sign and one fix
$5a month for Pro
free to start, no card
What people come here to get reasoned through

If you typed one of these into Google at midnight, this is the page.

  1. how do I make a big decision without missing something importantDecision Making Framework
  2. imagine my project already failed and work backward from therePre-Mortem Analysis
  3. who's right in this argument, get an unbiased verdictSettle the Argument (Neutral Referee)
  4. build the strongest case against something I already believeDevil's Advocate Analysis
  5. question a belief until I hit what it's actually built onSocratic Method Framework
  6. stress-test a theory's assumptions and the evidence behind itTheoretical Critique
  7. find the holes in my own thinking before someone else doesFind The Holes In My Thinking About This Idea
  8. what does this idea actually have to assume to be trueFind What Has To Be True For This Idea To Hold

Eight of sixty-six. Deciding, stress-testing, arguing both sides, and finding out what you actually believe and why.

Three real ones

What you type in about the real situation. What comes back reasoned through it.

Made the way you would make them: the real numbers, the real constraints, run once on the AI, not cleaned up after. The names and the businesses are invented. The reasoning is not.

Weighing a stable logistics job against a 12-person startup's first ops-hire offer

5 blanks · about 4 minutes · one run
You fill in
DecisionWhether to leave my stable job at a 200-person logistics company to join a 12-person startup as their first ops hire
Decision TypeCareer or professional
Decision StakesHigh, difficult to reverse, affects others
Options ConsideredStay and push for the Q1 promotion, take the startup offer, or negotiate a 3-month contract trial first
Time PressureThe startup needs an answer in 10 days
Your AI writes

What you're actually deciding

You must decide whether to stay at your current logistics firm (≈200 employees) and push for the Q-1 promotion that's been promised, jump now into a 12-person startup as its first operations hire, or try to carve out a short-term, three-month trial contract with the startup before making a full commitment.

The recommendation

Take the three-month paid trial (Option C) – if you can lock it down; otherwise, negotiate a part-time/consulting bridge (Option D).

Why

  • Information value: A short, bounded experiment dramatically reduces uncertainty about both the founders' competence and the fit of an ops-lead role.
  • Leverage: You can still press for the Q1 promotion at your current firm during or after the trial; having a concrete "outside test" strengthens your bargaining position.
  • Exit clarity: With predefined success/failure criteria, you avoid being stuck in a vague "pilot" that drags on.

Plus the full seven-option comparison table scored across financial security, growth, stability, impact, work-life, reputation, and payoff horizon, the cognitive biases most likely to distort this exact call (anchoring, loss aversion, the 10-day deadline's urgency effect), and what you'd need to believe for each option to be the right one.

How it works

Three steps. No blank page.

01

Pick the reasoning you need

A decision to make, a plan to pressure-test before it costs you, or an argument two people keep having. Search it the way you'd actually say it.

02

Feed it the real situation

Four to nine blanks: the options you're weighing, the constraints in play, both sides of the case. Not a blank page, your actual facts.

03

Send it to your AI, read what it flags

Ultra Prompt builds the expert prompt and hands it to ChatGPT, Claude, or Gemini. Anything it can't verify from what you gave it comes back marked, never guessed.

Past the template: reasoning that remembers what you already decided

Templates reason through one thing. An agent can hold the whole decision.

The same facts you type into a template can live inside an AI agent that knows your situation, your constraints, and the decisions you've already made, so every pressure-test and every verdict starts from your actual record instead of a blank chat. Three agents worth building:

Deciding

The Options Desk

Holds every option you're weighing and the reasoning behind each one. Name a new option or a new constraint; it re-runs the comparison instead of making you start over.

Runs in Claude Projects, a Custom GPT, or a local model
Pressure-testing

The Pre-Mortem Partner

Knows your project's real budget, timeline, and the people involved. Ask it to imagine failure at any horizon and it grounds every scenario in the constraints you already gave it, not generic risk.

Runs in your phone, through the AI you already use
Settling it

The Neutral Referee

Remembers the recurring argument, not just today's round. Bring it the same kitchen-table dispute again and it names the pattern instead of re-litigating from zero.

Runs in Gemini, ChatGPT, or Claude
Agent Studio

Build your options desk in an afternoon.

Agent Studio is in every Pro seat. Give it the decision you're facing, the constraints that aren't going away, and the way you actually think things through. Then bring it a new option or a new twist; back comes the comparison, the pre-mortem, or the verdict, grounded in what you told it, 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 Studio
Custom build

Or have it built around your decision.

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 build

Files are yours to keep. The library, updates, and support ride the subscription.

Pricing

One plan. Honest about what it does.

Pro
$5a month

Unlimited generates, all 66 thinking templates and the full 1,000+ library, every technique and pipeline, and Agent Studio.

Start with 10 free prompts

Free to start: 10 finished prompts a month, no card. Cancel anytime.

Agent Studio
Includedwith Pro

Build an agent that knows your situation and your constraints, then export it to Claude Projects, a Custom GPT, Gemini Gems, Copilot, or a local model. Your decision, on your own account.

See Agent Studio
Straight answers

The questions you ask before you trust a tool with a real decision.

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 tell me what I want to hear?

No. The templates ask it to flag the biases most likely to distort this exact decision and to state what you'd need to believe for each option to be right, including the one you're leaning toward. Read the caveats before you read the recommendation.

How is this different from just asking ChatGPT?

You can ask ChatGPT to help you decide and get a vague pro-con list. These templates ask for the stakes, the deadline, the real constraints, and the options you've already ruled out, then structure the reasoning around your specific situation instead of a generic framework.

Will it make up facts to fill gaps in what I gave it?

It's built not to. The pre-mortem and the neutral referee both mark anything they can't verify from your inputs instead of inventing a number, a study, or a statistic to sound complete.

Is this financial, legal, or medical advice?

No. It reasons through the situation you describe. A decision with real money, legal, or health stakes still deserves a professional's eyes before you act.

What happens to the decision after I make it?

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 the decision itself.

The deadline is real. Reason through it before it reasons for you.