The Gaps AI Won't Fix Without Your Input

Every business owner I talk to right now is asking the same question: "Should we be using AI for this?"

It's the wrong first question.

The right one is: "What's actually broken in how we work — and would AI make it faster, or just make it break faster?"

Optimization isn't a tool problem, it's a process problem

After years spent inside complex organizational systems — and now helping small businesses fix theirs — I've noticed a pattern. Companies buy an AI tool, plug it into a broken workflow, and expect the tool to absorb the dysfunction. It doesn't. It amplifies it.

If your intake process is inconsistent, an AI chatbot won't standardize it — it'll just handle the inconsistency faster and at a larger scale. If your team doesn't have clear documentation for how a task should be done, an AI assistant can't infer the "right way" — it'll guess, and sometimes guess wrong, at machine speed.

AI is a force multiplier. It multiplies whatever you already have — including the gaps.

The three gaps AI exposes fastest

• Undocumented processes. If the "how" only lives in one person's head, AI has nothing to learn from and nothing reliable to automate.

• Inconsistent decision-making. AI needs rules or patterns to follow. If two employees handle the same situation two different ways, the AI will too — unpredictably.

• No feedback loop. Businesses that optimize well treat every dropped ball as data. Businesses that don't, repeat the same failure with a shinier tool attached to it.

Where most of my clients already have the tool — they just aren't using it

Here's the part that surprises people: most small firms I work with already own the AI they're asking about. If you're on Microsoft 365, Copilot is now built directly into the everyday tools your team already opens — Outlook, Word, Excel, Teams. As of this month, it's part of the standard business plans, not a separate purchase most firms skip.

A few places I see it create immediate, unglamorous value once the underlying process is actually sound:

• Outlook — inbox triage and drafting. For firms fielding a high volume of client and case-related email, Copilot can summarize long threads and draft first-pass responses. This only works well once you've defined what a good response looks like — which loops back to needing the process documented first.

• Teams — meeting notes that actually get written down. Client intake calls, staff meetings, case strategy sessions — Copilot can generate a summary and action items automatically. For firms where "who owns the follow-up" tends to fall through the cracks, this alone closes a real gap.

• Excel — faster analysis, not just formulas. Instead of building a pivot table from scratch, you can ask Copilot to summarize trends in a data set — caseload by attorney, referral sources, intake response times — in plain language, in minutes instead of hours.

• Power Automate — the quiet workhorse. This is where actual optimization happens: automatically routing a new client intake form to the right person, flagging a file that's sat untouched for too long, or triggering a follow-up reminder. It's not flashy, but it's the piece that turns "we should really have a system for this" into an actual system.

None of this works if it's bolted onto a broken process. A firm with an inconsistent intake process that adds Copilot to Outlook just gets inconsistent responses faster. The order still ma

tters:

1. Find the gap.

2. Fix the process.

3. Then layer in the AI you likely already have access to.

This is the work I do

Process improvement, workflow audits, and AI-powered systems built for how a business actually operates — including getting real, practical use out of the Microsoft 365 tools most firms already pay for but rarely use to their full potential. If you've been wondering whether AI could help your business but suspect the real issue is upstream of that question, that's usually the right instinct.