2026-06-14 Jake Botticello

"We Need AI!" — What Your Business Actually Needs Instead

"We need AI" is a panic, not a plan. The businesses winning with AI didn't buy AI — they fixed a workflow that happened to need automation. Here's how to tell the difference before you spend a dollar.

Somewhere in the last year, “we need AI” became a thing business owners say to themselves at 11pm. A competitor mentioned it on a podcast. A board member asked what your AI strategy is. LinkedIn is wall-to-wall with people who’ve apparently automated their entire company and now work four hours a week from a beach. And so the sentence forms: “we need AI.”

Here’s the uncomfortable thing about that sentence: it’s not a plan. It’s a panic. And it’s the single most reliable predictor of wasting money on AI that exists.

Because “we need AI” is an answer in search of a question. It names a tool before anyone’s named a problem. And when you start from the tool, you end up exactly where 95% of enterprise AI pilots end up — with a thing you bought, bolted onto a process nobody examined, delivering no measurable impact, getting quietly abandoned before renewal. The technology didn’t fail. The sentence did.

This article is about the better sentence.

In the Article:

Why “We Need AI” Is the Wrong Sentence

Try this with any other tool and hear how absurd it sounds. “We need a forklift.” Do you? For what? Are you moving pallets? How many, how often, how heavy, and what’s it costing you to move them by hand right now? Nobody walks into a warehouse and announces “we need a forklift” without a load to lift. But that’s precisely what “we need AI” does — it announces the purchase before anyone’s identified the load.

The sentence feels productive because it feels like action. It feels like you’re keeping up. But it skips the only step that determines whether any of this works: figuring out what’s actually slow, expensive, or broken about how the work moves through your business right now. AI is a tool — a genuinely remarkable one — but a tool is only ever as good as the job you’ve correctly identified for it. Start from “we need AI” and you’ll buy the most impressive forklift in the showroom and discover you needed a hand truck.

Worse: AI bolted onto a broken process doesn’t fix the process. It automates it. If your onboarding is a mess, AI gives you a faster, more confident mess at higher volume. The chaos doesn’t go away — it scales.

What You Actually Need: A Workflow That Happens to Need Automation

Here’s the reframe that changes everything, and it costs nothing to adopt: you don’t need AI. You need a specific workflow to stop costing you so much, and some of those workflows happen to be ones AI is the right tool for.

That sentence does real work. It puts the problem first and the tool second, where they belong. It forces you to find the load before you buy the forklift. And it quietly reframes the whole project from “adopt a technology” — vague, trendy, unmeasurable — to “fix this expensive thing” — specific, gradable, real.

When you start from the workflow, three things happen automatically that never happen when you start from the tool. You discover what’s actually broken (often not what you assumed). You get a baseline you can measure against later, so you’ll actually know if the fix worked. And — frequently — you discover the best fix isn’t AI at all. It’s a process change. An integration between two tools you already pay for. A template. A clear owner for a step that had none. The AI-shaped solution is sometimes the answer; it is never the starting assumption.

The Three Questions That Replace the Panic

When the “we need AI” feeling hits, run these three questions instead. They take an afternoon and they save you from the expensive version of finding out.

1. What is the single most expensive way our work currently moves? Not the most annoying — the most expensive, in hours, errors, delays, or deals lost. Quote turnaround taking six days. Onboarding docs reworked one in five times. Status updates eating an hour of everyone’s morning. Find the load before you shop for the lift.

2. Why is it expensive — is it the process, or the volume? This is the fork. If the process itself is broken — unclear, undocumented, no owner — then no tool fixes it; you fix the process first, and maybe automate the result. If the process is sound but the volume is crushing you — same task, correctly done, just too many times for humans to keep up — that’s where automation, and sometimes AI specifically, earns its keep. Most “we need AI” panics are actually process problems wearing a technology costume.

3. What would “fixed” measurably look like? If you can’t name the number that would move — turnaround from six days to one, rework from 20% to 5% — you’re not ready to buy anything, because you won’t be able to tell if it worked. A fix you can’t measure is a purchase you can’t justify at renewal.

Answer those three honestly and the right move names itself. Sometimes it’s AI. Often it’s something cheaper, simpler, and already in the building.

“AI” vs. “Automation” vs. “The Right Tool” — Stop Conflating Them

Half the confusion in this whole conversation is vocabulary. People say “AI” when they mean three different things, and the conflation is what gets them sold the wrong thing.

Automation is making a defined, repeatable process run without a human pushing every button — the invoice that generates and sends itself, the lead that routes to the right rep automatically, the report that builds on a schedule. Most of it isn’t AI at all. It’s logic, integrations, and triggers. It’s also, for most small businesses, where the biggest immediate wins live — and it’s boring, which is why nobody’s tweeting about it.

AI is the subset of tools that handle the work automation can’t define cleanly in advance — drafting from context, summarizing messy input, answering open-ended questions, reading a document and pulling out what matters. It’s genuinely powerful for the fuzzy, judgment-flavored work. It is overkill, and often worse, for work that plain automation already handles deterministically.

The right tool is whichever of those actually fits the job you identified in the three questions. Sometimes that’s an AI agent. Frequently it’s an automation with zero AI in it. Occasionally it’s neither — it’s a process fix and a clear owner. The entire skill is matching the tool to the job, which you can only do after you’ve named the job.

“We need AI” collapses all three into one panicked purchase. “We need to fix how quotes move, and the right tool for the volume problem turns out to be an AI drafting step wired into our existing CRM” — that’s a plan. Longer sentence. Much cheaper outcome.

What the Winners Actually Bought

The businesses winning with AI didn’t buy AI. McKinsey found the high performers were nearly three times as likely to have fundamentally redesigned their workflows — meaning they started from the work, fixed how it moved, and brought in AI as one component of a redesigned process. They bought a solution to a named problem. The AI was an ingredient, not the meal.

The ones who lost bought “AI” — the meal as advertised on the box — bolted it onto an unexamined process, and waited for the magic. 91% of small businesses using AI report revenue gains; the gap between them and the abandonment pile isn’t budget or model quality. It’s whether they started from the problem or the tool. That’s the whole game.

The Better Sentence

So the next time “we need AI” forms at 11pm, replace it with the sentence that actually works: “What’s the most expensive way our work moves, why is it expensive, and what would fixed look like?”

Answer that, and you’ll either find a workflow worth automating — at which point AI might be exactly the right tool, applied with a target and a yardstick — or you’ll find a process worth fixing, and save yourself a subscription you’d have abandoned by spring. Both are wins. Both start from the work, not the tool.

That’s the difference between the 5% and the 95%. Not the AI. The sentence.

If you’d rather have someone run that diagnosis with you — find the expensive workflows, decide what actually needs automating, and build the right tool wired into how you really work — that’s exactly what we do at Pyris. We start from your work, not from a product we’re trying to sell you. Sometimes the answer is a custom AI agent. Sometimes it’s an automation with no AI in it at all. Either way, you’ll know the number it’s supposed to move, and whether it moved — and once it’s running, a two-page oversight register keeps it honest as your business keeps changing.

Take our AI Readiness Assessment for an honest read on where you stand, or book a no-cost discovery call — 20 minutes, no pitch deck, no pressure. Bring us the panic; we’ll help you turn it into a plan.

Jake Botticello is the founder of Pyris Consulting, where he and his team build custom processes, integrated systems, and purpose-built AI automations for founder-led service businesses. He starts from the work, not the tool — every time.

Frequently Asked Questions

Why do most AI projects fail?

Most AI projects fail because they start from the tool instead of the problem. "We need AI" names a tool before anyone's named a problem — so the tool gets bolted onto a process nobody examined, delivers no measurable impact, and gets quietly abandoned before renewal. The technology didn't fail. The sentence did. Roughly 95% of enterprise AI pilots end up exactly there.

What's the difference between AI and automation?

Automation makes a defined, repeatable process run without a human pushing every button — the invoice that goes out on its own, the follow-up that triggers on a form submission, the report that builds itself on a schedule. AI is the subset of tools that handle the work automation can't define cleanly in advance — drafting, reasoning, pattern recognition, handling variation. Automation is where the biggest immediate wins live, and it's boring, which is why nobody's tweeting about it. AI is overkill — and often worse — for work plain automation already handles.

Does my small business actually need AI?

Probably not the way the panic implies. You don't need AI — you need a specific workflow to stop costing you so much, and some of those workflows happen to need automation to fix them. Start from the expensive workflow, not the tool. Frequently the best fix isn't AI at all: it's a process change, a cleaner handoff, or a single source of truth that didn't require a subscription.

How do I know whether I need AI or automation?

Ask why the workflow is expensive: is it the process, or the volume? If the process itself is broken — unclear, undocumented, no owner — no tool fixes it, because tools automate what's there and chaos automated is just faster chaos. Fix the process first. If the process is sound but the volume is crushing you — the same task, done correctly, just too many times for humans to keep up — that's where automation, and sometimes AI specifically, earns its keep. Most "we need AI" panics are volume problems wearing a complexity costume.

Will AI fix a broken business process?

No — it makes a broken process faster. AI bolted onto a broken process doesn't fix the process; it automates it. If your onboarding is a mess, AI gives you a faster, more confident mess at higher volume. You've bought an escalator going the wrong direction. Fix the process first, then decide whether it's worth automating.

What questions should I ask before buying an AI tool?

Three: (1) What is the single most expensive way our work currently moves — in hours, errors, delays, or deals lost? (2) Why is it expensive — the process, or the volume? (3) What would "fixed" measurably look like — turnaround from six days to one, rework from 20% to under 5%? Answer those three honestly and the right move names itself. Sometimes it's AI. Often it's something simpler. Occasionally it's nothing.

What did companies that succeeded with AI actually do differently?

They started from the work, not the tool. McKinsey found high performers were nearly three times as likely to have fundamentally redesigned their workflows before or during AI adoption — they fixed how the work moved and brought AI in as one component of a redesigned process. They bought a solution to a named problem; the AI was an ingredient, not the meal. The gap between the winners and the abandonment pile isn't budget or model quality. It's whether they started from the problem or the tool.