PP
Home/Blog/When AI Is Overkill: Simple Solutions for Simple Problems
AI
15 January 2026
5 min read

When AI Is Overkill: Simple Solutions for Simple Problems

Not every problem needs an AI agent. Here's how to tell when a spreadsheet, a Zapier flow, or a simple script will do the job better, and when AI is genuinely the right call.

AI StrategyAutomationBusinessProduct Strategy
AG

Hey, my name is Anthony. I started Product In Your Pocket to help people build software that works. I hope you enjoy this read. Reach out to me on LinkedIn or contact us if you have any questions.

I build AI products for a living. I'm about to talk you out of using AI.

This might seem counterintuitive. I sell AI strategy, AI agents, and AI automation. But the fastest way to lose a client's trust is to recommend AI for a problem that doesn't need it.

The best product engineers recommend the right tool, not the most impressive one. Sometimes that tool is a spreadsheet.

The AI hype cycle is real

Every business owner I talk to right now wants "AI something." AI customer support. AI analytics. AI content. AI everything.

When I ask what specific problem they're trying to solve, the answer is often something like:

  • "We spend too much time copying data between systems"
  • "Our team forgets to follow up with leads"
  • "We can't keep track of our inventory"

These are real problems. They deserve real solutions. But most of them don't need AI.

When you don't need AI

Problem: "We copy data between systems manually"

You don't need: An AI agent that reads data from one system and writes it to another.

You need: Zapier, Make, or n8n. A simple automation that triggers when data changes in System A and updates System B. No intelligence required. Just plumbing.

Cost comparison: AI agent: $3,000-5,000. Zapier workflow: $0-50/month.

Problem: "We forget to follow up with leads"

You don't need: An AI that analyses lead behaviour and predicts optimal follow-up timing.

You need: A CRM with automated reminders. Set a rule: "If no response in 3 days, send a follow-up email." Done.

Cost comparison: AI system: $4,000-8,000. CRM automation: built into the tools you're already paying for.

Problem: "We can't track our inventory"

You don't need: An AI-powered demand forecasting system.

You need: A shared spreadsheet or a simple inventory app. For most small businesses, knowing what you have is the problem, not predicting what you'll need.

Cost comparison: AI forecasting: $5,000-12,000. Airtable or Google Sheets: free.

Problem: "We need to categorise support tickets"

You don't need: An LLM that reads every ticket and classifies it with natural language understanding.

You need: Three categories and a dropdown menu. If your support team is handling 20 tickets a day, a human can categorise them in seconds. The bottleneck isn't classification. It's resolution.

Cost comparison: AI classifier: $2,000-4,000. Dropdown menu: already exists in your help desk.

Problem: "We want to send personalised emails"

You don't need: An AI that generates unique email copy for every customer.

You need: Three email templates with merge tags. "Hi {first_name}, since you bought {product}, you might like {related_product}." That's not AI. That's mail merge.

Cost comparison: AI email system: $3,000-6,000. Mailchimp with segments: $20/month.

When you actually need AI

AI earns its keep when the problem involves:

Unstructured data

If you're dealing with free-text reviews, images, audio, or documents that don't follow a consistent format, AI is the right tool. A human can read 10 reviews. AI can read 10,000 and find patterns.

Example: Analysing hundreds of product reviews to identify the top 5 complaints. No spreadsheet formula can do this.

Complex decision-making with many variables

When the decision requires weighing dozens of factors simultaneously and the "right answer" isn't obvious from a simple rule.

Example: A running shoe recommendation engine that considers foot shape, running style, terrain, injury history, and personal preferences. That's genuinely complex, not a lookup table.

Scale that humans can't match

When the volume of work is physically impossible for humans to handle in the required timeframe.

Example: Processing 500 job applications and surfacing the top 20 candidates based on nuanced criteria. A recruiter can't read 500 CVs in a day. An AI agent can.

Conversations that need to feel natural

When you need to interact with users in natural language and handle unpredictable inputs.

Example: A customer support agent that handles returns, answers product questions, and escalates complex issues, available 24/7 across time zones.

The decision framework

Before building anything with AI, ask these three questions:

1. Can I solve this with a rule? If the logic is "if X then Y," you don't need AI. Use an automation tool, a formula, or a simple script.

2. Can I solve this with a template? If the output follows a predictable pattern, use a template with variables. Don't generate what you can fill in.

3. Does this require understanding? If the solution needs to interpret meaning, handle ambiguity, or process unstructured information, now you're in AI territory.

Why this matters for your budget

Every dollar you spend on AI for a problem that didn't need it is a dollar you can't spend on a problem that does. I'd rather build you one excellent AI agent that transforms your business than five mediocre ones that could've been Zapier workflows.

The businesses getting the best ROI from AI aren't the ones using it everywhere. They're the ones using it precisely, on the problems where nothing else works.

What I recommend

If you're not sure whether your problem needs AI, start with the simplest solution. Try the spreadsheet. Try the automation. Try the template. If those solutions crack under the weight of your actual needs, then you have a clear case for AI, and you'll know exactly what you need it to do.

Book a free consultation and I'll tell you honestly whether your problem needs AI or a spreadsheet. No judgment either way.

About us

We turn your goals into AI and software that actually works

A team of product engineers based in Queenstown, NZ. We work with you to understand the problem first, then build the right thing — not just the possible thing.

Book a consultation