The Thinking Partner: How to Use AI to Make Better Decisions

Here's what the headlines say: "AI Will Replace Your Job" or "ChatGPT Makes All Your Decisions."

Here's what's actually happening: a growing number of professionals are quietly using AI as a thinking partner—not to automate their judgment, but to sharpen it.

And they're not telling you about it because it sounds either trivial ("I use ChatGPT") or crazy ("I argued with a bot for three hours about my pricing strategy").

But it's working. Let me show you how.

A CFO at a mid-sized company had a problem. His competitor had better pricing power, and his team couldn't figure out why. The usual options—hire consultants, commission reports, schedule endless meetings—would take weeks and cost a fortune.

So over a weekend, he tried something different.

He fed his company's financials and publicly available data about the competitor into ChatGPT. Then he asked: "Why does this company have better pricing leverage than we do?"

The AI didn't just regurgitate the data. It started asking questions his team hadn't thought to ask. It spotted patterns in the competitor's customer concentration that his analysts had missed. It connected dots between their pricing structure and their cost of capital that hadn't been obvious.

By Monday morning, he had a hypothesis worth testing—and he'd spent exactly zero dollars and about four hours.

The AI didn't replace his judgment. It made his judgment better.

If you tried ChatGPT in 2023, you might have been underwhelmed. It was good at writing emails and summarizing articles, but it wasn't great at thinking.

That's changed.

The newest AI models—Claude's Opus, Google's Gemini 3.0—don't just retrieve information. They reason. They break down complex problems step by step, consider multiple angles, and explain their logic.

This matters because it means AI can now be wrong in interesting ways. When you ask it to challenge your thinking, it doesn't generate generic objections. It finds the specific weak point in your argument, based on your data.

That's what makes it useful.

Forget the buzzwords. Here's what's working in practice:

1. Pre-mortems before decisions

Before committing to a strategy, leaders are asking AI: "I'm about to do X. Assume it fails spectacularly. What went wrong?"

This is a pre-mortem—a technique from psychology where you imagine failure and work backward. It's brutally effective at surfacing risks, but most teams are too polite or too optimistic to do it well.

AI doesn't care about your feelings.

I talked to a marketing director who runs every campaign through this filter now. The AI keeps finding the thing she doesn't want to think about—the budget assumption that's too rosy, the competitor response she's pretending won't happen. "It's annoying," she told me. "It's also usually right."

2. Devil's advocate on demand

There's an old story about Alfred Sloan at General Motors. At a meeting of executives, Sloan proposed a major strategic decision. When he asked for feedback, everyone agreed.

Sloan postponed the decision.

"Either you don't know enough to point out the downsides," he said, "or you're afraid to disagree with me. Come back when someone can tell me why this is a bad idea."

Most of us don't have the courage to do this. And even if we did, our teams don't have the courage to push back.

Here's the thing: some companies have figured out how to build this discipline into their processes. At Amazon, when a hiring manager wants to promote someone on their team, they have to write the strongest case against the promotion—in the same document where they're arguing for it. It's uncomfortable. It's also incredibly clarifying.

Most companies don't have this built in. But you can create your own version with AI.

The prompt is simple: "I'm planning to [decision]. You're a skeptical board member who thinks this is a mistake. Make your case."

What comes back isn't generic criticism. It's specific, logical pushback based on the context you provided. And because it's AI, there's no ego involved. You can argue back, refine your thinking, and move forward—without the political minefield of actual disagreement.

3. Scenario testing at speed

Before AI, scenario planning meant spreadsheets, assumptions documents, and days of work. Now you can do it in minutes.

"What if we raise prices 5% and our main competitor matches us?"

"What if we raise prices 5% and they don't match us?"

"What if we raise prices 8% but only on our premium tier?"

AI can model these scenarios faster than you can schedule a meeting to discuss them. The outputs aren't perfect—AI doesn't know your market like you do—but they're good enough to spot the obvious landmines before you step on them.

One CEO described it as "having a really smart intern who works 24/7 and never gets tired of your questions."

Now the less fun part: AI can also make you dumber.

Harvard Business Review ran an experiment where executives predicted stock prices. Half used ChatGPT. Half discussed with peers.

The ChatGPT group became more confident, more optimistic, and made worse predictions.

Why? Because AI sounds authoritative. It gives detailed answers without hedging or second-guessing. And when something sounds that confident, we stop questioning it.

This is the trap: AI gives you the feeling of rigorous analysis without the rigor. If you treat it like an oracle instead of a sparring partner, you'll make worse decisions while feeling better about them.

The solution is simple but requires discipline: never accept AI output at face value. Push back. Ask "why." Demand different perspectives. Treat it like that brilliant but inexperienced colleague—useful, but not infallible.

The best way to think about AI in decision-making: it's a new hire on your team.

You wouldn't let a new hire make major decisions alone. But you also wouldn't ignore their analysis just because they're new.

You'd review their work. Push back on assumptions. Use their output as input to your judgment.

That's the relationship that works.

AI won't make you a better decision-maker by itself. But if you know how to interrogate it, challenge it, and extract what's useful, it's the best thinking partner you've never had to pay.

Just remember: the AI is the new hire. You're still the one accountable for the decision.

You're going to hear this term everywhere in 2026. Here's what it means:

Traditional AI is reactive—you ask, it answers. Agentic AI can plan, act, and learn on its own toward a goal you set.

Think of it like the difference between a search engine and a research assistant. A search engine waits for you to type a query. A research assistant figures out what to look up, what to ignore, and when to come back with findings.

Gartner predicts that by 2028, 15% of day-to-day work decisions will be made autonomously by agentic AI—up from essentially zero today.

Drop this at your next meeting: "The shift from reactive AI to agentic AI is like going from calculators to autopilot—you're still responsible, but the machine is doing more of the work."

Here's a simple protocol you can use this week to test AI as a thinking partner:

Pick a decision you're facing—anything from "Should we hire another person?" to "Should we launch this product?"

Then run it through this three-step process:

Step 1: The Setup Open ChatGPT or Claude and describe your decision in 2-3 sentences. Include the key factors you're weighing.

Step 2: The Challenge Ask: "You're a skeptical advisor who thinks I'm missing something important. What are three considerations I'm overlooking?"

Step 3: The Opposition Ask: "Now argue the opposite position. What's the strongest case against the direction I'm leaning?"

Bonus Step: Second-Order Thinking Ask: "If I go ahead with this decision, what are the second and third-order effects I might not see right now? What changes six months down the line?"

This is where AI really shines—connecting dots you haven't connected yet.

Don't just read the answers—actually think about them. Do they change your view? Do they surface something you genuinely hadn't considered?

Most people who try this are surprised at least once. That's the value.

McKinsey's "AI in the Workplace" report (January 2025) tracks how organizations are actually deploying AI for decision-making—not the hype, just the results. If you're serious about integrating AI into your workflow, start here.

That's Gist for this week. See you next week.