Somewhere in the last 18 months, “Which AI should I use?” became the most common question I get.
From clients. From workshop participants. From people at networking events who find out I do AI consulting and immediately want the answer to the one question they've been sitting on.
And I get it. The options are genuinely overwhelming. ChatGPT, Claude, Gemini, Grok, Perplexity — each one with its own pricing tiers, quirks, and use cases. Picking one and sticking to it feels like the safe move.
But here's the thing: tool loyalty is beginner mode.
The people I see getting outsized results from AI have stopped asking which tool to use. They're asking a different question: which tool wins at this specific job?
That's the shift. And it changes everything about how you use AI.
The Cruise Ship, the Speedboat, and the Yacht
Here's an analogy I used in a coaching session recently that clicked immediately for my client.
Think of Google (Gemini) as a massive cruise ship. It has the most data, incredible underlying tech, and it's especially good at visual work — images, video, anything that needs real-world grounding. But it takes more prompting to steer precisely.
Claude is the speedboat. It's a startup that moves fast. Technical reasoning, complex coding, working through a multi-step problem systematically — Claude is genuinely faster and more precise here than the others. The integrations with Google Drive and Gmail make it a strong choice if you're in a Google-native workflow.
ChatGPT is the reliable yacht. Everyone knows it. It's your daily driver — general questions, brainstorming, first drafts. It's not always the best at any one thing, but it's consistently good at almost everything and the interface is intuitive.
Three tools. Three distinct personalities. None of them is the “right” answer in isolation.
What Multi-Tool Fluency Looks Like in Practice
During a Lindy fireside chat a few months back, I talked about how I actually use these tools day-to-day. I don't wake up and think “today I'm a ChatGPT guy.” I think about the job first.
Writing a first draft of a proposal? ChatGPT. It's fast, good at tone, and I can iterate quickly.
Debugging why an agent workflow keeps breaking? Claude. It walks through reasoning in a way that actually helps me find the problem instead of just giving me a new answer that's also wrong.
Creating visual assets or summarizing a 50-page document? Gemini. Specifically Gemini 3.0 Pro right now — for condensing large datasets and multi-step research tasks, it's the best I've used.
This is what I teach in my workshops as moving from AI Assisted to AI Workflows — you're not just using AI for one-off tasks, you're routing work deliberately based on what each tool does best.
The Multi-Model Agent
Here's where it gets interesting for people building AI agents.
I've been building agents where different steps of the same workflow run on different models. The research step might use a cheaper model like Claude Haiku — it's fast, cost-effective, and I don't need heavy reasoning for simple information gathering. Then the summarization step switches to Gemini 3.0 Pro because it handles condensing large amounts of information better than anything else right now.
Same agent. Three different models. Each one doing what it's best at.
The economics work out too. Haiku costs fractions of a cent per task. Gemini 3.0 Pro for summarization gives you better output than paying for the premium tier on a model that's not as good at that specific job.
I used to think this was overkill. Now I think it's just… obvious. You wouldn't use the same tool for every step in a kitchen. You use a knife for cutting, a pan for cooking, a blender for blending. AI is the same.
How to Actually Build This Skill
If you're reading this and thinking “okay but where do I start” — here's the honest answer.
You don't start by subscribing to everything at once. That's expensive and overwhelming.
You start with one daily driver and you hit its limits.
Most people start with ChatGPT. Use it for 30 days and pay attention to where it frustrates you. Where does it give you vague answers? Where does it fail on technical tasks? Where does the output feel generic?
Those frustration points are your entry points for the other tools.
When you hit a wall on a technical problem, open Claude and try the same prompt. Feel the difference in how it reasons through things.
When you start working on anything visual, create a Gemini account and try it there. Compare the outputs.
You grow into multi-tool fluency the same way you grew into your current software stack — through repeated use, noticing what works, and expanding when you hit a ceiling.
The goal isn't to use every AI tool. The goal is to stop using the wrong tool for the job.
The Shift That Changes Everything
I've watched a lot of people make this transition — from picking one tool and cramming everything through it, to routing work based on the job at hand.
The biggest indicator of where someone is: do they ask “which AI is best?” or do they ask “what's this tool best at?”
The first question assumes there's a winner. The second question assumes each tool has a role.
One question keeps you stuck at one tool. The other one makes you 10x more effective across all of them.
If you're ready to build a proper AI toolkit and learn how to route work across models, that's a big part of what we cover in my AI workshops. One day, hands-on, and you walk out with a system that actually fits your workflow — not a generic list of prompts.
But even without the workshop: pay attention to where your current tool frustrates you. That's your next tool waiting to be discovered.
