OpenAI drops AgentKit. Anthropic releases Claude with computer use. Google launches their Agent Development Kit. Microsoft announces yet another AI framework.
And somewhere in the middle of all this, you're supposed to figure out which tools to use, what automations actually save money, and whether that shiny new AI feature is worth building.
It's a mess. A proper Wild West where everyone's claiming to have struck gold.
The Real Problem Isn't Lack of Tools
I've learned a lot while building AI automation tools—sales training platforms, chatbots, workflow integrations:
The problem isn't that we can't build things. The problem is we can build almost anything.
With AI now, the technical barriers have basically disappeared. Want a chatbot? Twenty minutes. Need to analyze sales calls? An afternoon. Want to automate your entire email workflow? Probably doable by the weekend.
But here's the question nobody's asking: Should you?
Not Every Problem Needs an AI Solution
I've seen companies spend thousands building AI tools that either:
- Solve problems they don't actually have
- Automate processes that weren't worth automating
- Add complexity where simplicity would work better
- Look impressive in demos but break in production
Just last month, I spoke with a startup that wanted to build an AI agent to manage their customer support. Sounds reasonable, right?
Except they had 10 customers. Ten.
They didn't need AI. They needed a spreadsheet and a phone.
The Questions You Should Ask First
Before you build (or pay someone to build) any AI automation:
1. What's the actual cost of NOT automating this?
If your team spends 2 hours a week on a task, and automation costs £5,000 to build and maintain... do the math. Unless those 2 hours are preventing something more valuable.
2. Will this break when things change?
AI tools can be brittle. Your perfectly-tuned chatbot might fall apart when your product changes, your pricing updates, or your team grows. What's the maintenance cost?
3. Are you solving a people problem with technology?
Sometimes the issue isn't the process—it's communication, training, or unclear goals. AI won't fix that. It'll just automate the dysfunction faster.
4. What's the simplest solution that could work?
Could you fix this with a better spreadsheet? A Slack channel? A weekly meeting? Don't overcomplicate things just because AI exists.
Where AI Actually Makes Sense
Don't get me wrong—there ARE incredible opportunities right now. The businesses winning with AI aren't the ones using every new tool. They're the ones being strategic about it.
AI excels when:
- Volume justifies automation: Processing hundreds of sales calls, support tickets, or data entries
- Speed matters: Real-time insights, instant responses, rapid analysis
- Personalization at scale: Customizing experiences for thousands of users
- Pattern recognition: Spotting trends humans would miss in large datasets
- Handling ambiguity: Natural language queries, unstructured data, adaptive responses
If your use case doesn't fit these categories, you might not need AI. And that's absolutely fine.
The Can vs. Should Framework
Here's how I approach it now:
Just because we CAN build it doesn't mean we SHOULD.
Before diving into any AI project, three questions:
- What's the actual business outcome we're after? (Not "cool AI tool"—real results)
- What's the simplest solution that achieves that outcome? (Maybe it's not AI)
- If AI is the right tool, what's the ROI timeline? (Be honest about costs)
Sometimes the answer is "build something custom with N8N and OpenAI."
Sometimes it's "use an off-the-shelf tool."
Sometimes it's "don't automate this at all."
Navigating the Noise
The AI Wild West isn't slowing down. If anything, it's getting faster.
More tools. More frameworks. More "you've got to try this" messages in your inbox.
But the winners won't be the businesses that use every new tool—they'll be the ones that know which tools to ignore.
The companies that ask "should we?" before "can we?"
The teams that automate strategically, not desperately.
What SHOULD You Build?
If you're trying to figure out where AI fits in your business—or whether it fits at all—let's have a chat.
I've spent the last few months neck-deep in AI automation: building sales training tools (think Gong, but bespoke), chatbots, and workflow integrations with N8N, OpenAI, and various other platforms. I've seen what works, what wastes money, and what looks impressive until it doesn't.
Here's what I'm offering: Let me take a look at your business and tell you what you SHOULD build. Then, if it makes sense, we CAN build it together.
No filler. No "let's add AI because AI is hot." Just honest advice about whether automation will actually move the needle for your business.
Drop me a message at dw@techconnex.co.uk or find me on LinkedIn, Slack, Discord, or pretty much any platform.
Let's figure out what's actually worth building.
A Quick Note on AI and Irony:
Yes, I used Claude to help write this post about AI automation. The irony isn't lost on me.
BUT - Claude has so much context on my projects, my voice, and how I think about these problems that it honestly sounds more like me than I do half the time. That's the point, isn't it?
AI isn't about replacing thinking - it's about amplifying it. I still had to know what I wanted to say, what advice actually matters, and what's worth building. Claude just helped me say it better and faster.
That's the difference between using AI well and using it badly. Know what you want to achieve, then use the tools to get there faster.