Lesson 3 of 7
Fast and cheap vs slow and smart
Inside every AI family there's a second choice that matters more than the brand name: a quick, cheap setting and a slower, cleverer one. Here's how to tell which a task needs.
Once you’ve picked a family, ChatGPT, Claude or Gemini, you meet a second choice that trips a lot of owners up. Inside that family there usually isn’t just one model. There are several, and they are not simply “better” and “worse”. They trade off speed and cost against depth of thinking, and knowing which end of that trade-off a job needs is what separates someone who gets real value from AI and someone who either overpays or under-delivers.
Think of it like staffing a job
You wouldn’t put your most experienced, most expensive person on a job that just needs someone to load the van. Equally, you wouldn’t hand your trickiest client negotiation to the newest apprentice. AI models work the same way.
Fast and cheap settings answer almost instantly and cost very little to run. They handle straightforward, repetitive work well: sorting, tagging, short replies, simple rewrites. What they give up is depth. Hand them something genuinely tricky and they are more likely to slip.
Slow and smart settings take longer and cost more per use, but they think more carefully, hold more detail in mind at once, and handle complex or high-stakes work far better.
A balanced middle setting is where most owners actually live day to day: quick enough, capable enough, sensibly priced for general use.
The exact names for these tiers change often enough that printing them here would date badly. What doesn’t change is the shape of the choice, and every major assistant offers some version of it.
When fast and cheap is exactly right
Reach for the quick, cheap setting when the job is simple and you’re doing a lot of it. A gutter clearing firm sorting a hundred inbound enquiries into “quote request”, “existing customer” and “spam” doesn’t need its most capable setting for that. Neither does a salon tagging a stack of Google reviews as positive or negative, or a firm pulling postcodes out of a long list of booking forms. Using the top-tier setting here is not more careful, it’s just slower and more expensive for no real benefit.
When you need the smart one
Reach for the slower, cleverer setting when the job is unusual, hard, or expensive to get wrong. Working through an awkward supplier contract and spotting the risks. Reasoning across a long report or a stack of site surveys to answer one specific question accurately. Planning a job with a lot of interlocking parts, like a full kitchen fit with three trades to coordinate. Anything a customer or a regulator will actually read, where a confidently wrong answer could cost you money or trust. Here, the extra care is the entire point. You’re not doing this a thousand times a day, so the higher cost barely registers, and the better judgement is worth it.
The habit worth building: route the job, not the tool
Stop thinking of “which setting” as something you choose once. Think of it per job. Default to the balanced middle for your general daily work. Drop to fast and cheap for anything high-volume and simple. Step up to slow and smart only for the occasional hard or high-stakes task. That single habit, applied without much thought once it’s a habit, keeps your quality high where it counts and your spending low everywhere else.
Next, we’ll get specific about which setting, and which kind of tool altogether, actually suits the jobs you do most: writing, analysis, customer replies and research.