Two kinds of decisions: some calls need a process, some need a second
Daniel Kahneman showed that we think in two modes, fast and slow. Business decisions split the same way, and until recently software could only help with the slow kind. What changes when AI tells you how sure it is.
A customer asks if you can come on Saturday. A lead emails at 11pm. A client suddenly goes quiet. Each one is a two-second call for you, and each one waits two days until you get to it. Most of your business isn't slow because the thinking is hard. It's slow because every quick judgment queues up behind you.
Ask any business owner how they make decisions and you'll get two different answers.
Some decisions get a process. Setting a price, hiring someone, signing a contract: you gather the facts and work through them carefully.
The other kind never gets a process, because there's no time for one. Is this lead serious? Is this complaint about to escalate? Has the client gone quiet on this deal? You just know, in a second, from experience. And you're usually right.
Fast and slow
The psychologist Daniel Kahneman described these two modes in Thinking, Fast and Slow. System 1 is fast, intuitive and automatic. System 2 is slow, deliberate and effortful. We use both constantly, and each is good at different things.
Software has always been a System 2 tool. Every automated decision had to be spelled out in advance as a rule: if this, then that. That's why the fast judgment calls in a business, the ones that pile up in the owner's inbox, stayed stuck with a person.
My twist: the answer is only useful if it comes with a confidence level
Newer AI models can now make fast, System 1-style judgments, like "this lead looks serious" or "this message is urgent". But fast answers are only safe if you know how much to trust each one.
That's the part many AI tools get wrong. They sound certain even when they aren't, so you end up supplying the doubt yourself, and a confident-sounding answer gets trusted more than it deserves.
The rule I build into every decision system is simple:
Act fast on the answers it's sure about. Stop and check the ones it isn't. Never let it hide the difference.
I've been building this for years
This isn't a new idea for me. It's how we built the original Mindreader product, and the lesson carries into the AI services work Mindreader is moving toward now.
Mindreader's original product predicts a person's communication style from their photo and from what they've written. Early on we decided the system had to say when it was guessing. It combined the two predictions under explicit rules and would only call a result "very confident" when the evidence was strong: enough text to read, and a clear gap between the top personality trait and the next one. A short sample was always marked "not very confident", however clear it looked. That way salespeople knew which readings to lean on and which to treat as a hint.
I use the same pattern today:
- Chatbots with a pass mark. Answers are scored out of 100, and only those scoring 90 or above reach the customer. The rest go back for revision or to a person. How that works.
- Lead and inbox tools that show their confidence. The Lead Fit Scorer and Inbox Money Filter on my Tools page display how sure they are and escalate the unsure cases, instead of pretending every call is equally solid.
Sorting your own decisions
Take the decisions that land on you every week and put them in three groups:
| Type | Examples | How AI can help |
|---|---|---|
| Slow and important | Pricing, hiring, contracts | Gathering facts and drafting options. You decide |
| Fast and frequent | Is this lead serious? Is this urgent? Who should handle this? | A fast judgment with a confidence score, acting on the confident cases |
| Fast but risky | Complaints, refunds, anything legal or safety-related | Flagging and routing to a person, fast. Never deciding alone |
The middle row is where most owners lose hours, and where confidence-scored AI helps most. The bottom row is where speed is the wrong goal.
Jeff Bezos makes a similar point about one-way and two-way doors in his 2015 Amazon shareholder letter: decisions you can easily reverse should be made fast, and ones you can't should be made carefully. Confidence scores let you apply that to hundreds of small calls a day, not just the big ones.