When AI tools started rolling out at pace, LinkedIn turned into a running show-and-tell. Look what I built. Look what I automated. Look at the role I just eliminated. And while some of it was genuinely interesting, most of it missed the point entirely.
AI is powerful. I use it every day across a wide range of work, and I've seen what it can do for lean teams operating in categories that used to require significantly more headcount. But it still requires a human on the other end of it. Judgment. Discernment. A clear sense of what problem you're actually trying to solve. Without that, you're not becoming more efficient. You're just producing more, faster, and adding to an internet that already has more content than anyone can consume.
That's not the argument I want to make today, though. Today I want to talk about overtooling.
The Tab Collector Problem
I've tested more than 300 AI tools. It's expensive, and at times it's more time-consuming than doing the actual work. That may sound like a strange admission from someone who built an AI tool directory, but it's exactly why I built it. Testing that many tools taught me something more useful than which ones work: it taught me how easy it is to accumulate tools that don't.
Most people collecting AI tools are doing it the same way people collect browser tabs. One for writing. One for design. One for video. One for automations. One for note-taking. One for transcription. One for SEO. Before long they're paying for five subscriptions that overlap, barely using three of them, and still feeling behind.
That's not efficiency. That's tool clutter.
Tool clutter creates its own kind of drag. More logins. More learning curves. More decisions about which tool to open for which task. More time spent evaluating whether the thing you're paying for is actually working. More money leaving the business without a clear return. The irony is that the productivity problem the tools were supposed to solve gets replaced by a different productivity problem, one that's harder to see because it looks like progress.
The Question That Should Come Before the Tool
Before adding anything to a workflow, it's worth slowing down enough to identify what's actually broken.
Where do you lose time each day, each week, each month? What are the tasks you avoid because they're tedious, not because they're hard? What skills are missing from your team, and is that gap causing a real bottleneck? Where does work slow down in ways that compound over time?
Those questions matter because the right tool should answer one of them directly. It should reduce friction that currently exists, fill a gap that's genuinely costing something, or free up time that's being spent on work a person shouldn't be doing. If a tool doesn't map to one of those things, it's adding complexity rather than removing it.
This is why I tell people to try before they buy and to be honest about what they find. Use the free version. Build the workflow. See whether it actually changes how the work gets done, not just whether the demo was impressive. Someone else's recommendation means they found a solution to their problem. That has no bearing on whether you have the same problem.
What a Tool That's Actually Working Feels Like
A good AI tool should feel like relief. Not novelty, not cleverness, not something worth posting about. Relief. The kind that comes from realizing you no longer have to spend two hours on something that used to take two hours.
It should help you move faster without lowering the quality of the output. It should reduce the mental overhead of repetitive tasks that consume time without creating value. It should earn its place in the workflow quickly, not after six months of trying to make it fit.
That's the filter worth applying. Not whether it's new. Not whether it has a waitlist. Not whether it's been covered in every newsletter this week. Whether it actually helps with something that currently isn't working well enough.
The tools that pass that test are worth paying for, learning, and integrating properly. The ones that don't are worth cutting, even if you already paid for them. Sunk cost is not a reason to keep something in a workflow that isn't delivering.
Fewer, Better
AI can absolutely make work better. I've seen it happen enough times to say that without qualification. But the efficiency gains come from choosing tools that solve real problems and using them well, not from accumulating everything that launches.
The smartest AI workflow I've encountered isn't the most elaborate one. It's usually the leanest one that still gets the job done. A handful of tools used consistently, integrated thoughtfully, and evaluated honestly against the problems they were brought in to solve.
More tools is not more efficient. Sometimes the most productive decision is the one where you don't add anything at all.
I built my AI resource directory with exactly this in mind. Each tool is organized by category and labeled as free, paid, or freemium so you can find what fits without wasting time on what doesn't. Explore it at resources.taneilcurrie.com