Operational & AI Consulting

Fix the process that is eating your week

Most owners who call want to "do something with AI" or "fix operations." When I ask which problem, the room goes quiet. This practice starts there. We map how the work actually moves, find where the time and money go, and only then decide whether the fix is a tool, a process change, or a security control. Where AI takes the first pass, a person reviews every output before it leaves the building.

Start with the Map

Who This Is For

Is this right for your business?

Small and mid-size businesses, and the MSPs that serve them, whose operations have not kept pace with their size. Usually the owner or one senior person has become the review desk for everything.

How It Works

How an engagement works

Four steps, always in this order. The order is what keeps you from buying a tool for a pain point you have not found yet.

Step 1

Map the process

How the work actually moves today, walked with the people who do it: who touches it, where it waits, what gets re-done, which tools it passes through. Not the org chart and not the procedure manual. The map is usually the first time the owner has seen the whole thing laid out end to end.

Output: a comprehensive map of the process, end to end: every step, who touches it, the systems and documents it passes through, where it waits, where it gets re-done, and what each stage costs.

Step 2

Find the pain points

The map shows where the time and money go. We pull out the specific points of friction, the ones the people doing the work agree on, and put a number on each: hours, errors, delays, or dollars, counted before anything changes. Sometimes that is a stopwatch and a week. If we cannot put a number on it, we are not ready to buy a fix for it.

Output: a ranked list of pain points with a baseline number on each, in units you already use.

Step 3

Decide the fix

A tool, a process change, a security control, or sometimes nothing. The measurement decides, not the vendor. Many problems turn out to be a hand-off or a missing rule, and the cheapest fix is a change to how the work moves, not new software.

Output: a written recommendation with the expected gain, the cost, and what we will measure after.

Step 4

Implement with a person reviewing every output

Build it, run it, and measure again against the baseline. Where AI or automation takes the first pass, a named person reviews every output before it reaches a client, a regulator, or the books. The tool drafts. Your people decide.

Output: the working process, the review rule written down, and the before-and-after number.

Published Examples

Three published examples of the order working

These are not Tech Cartographer client engagements. They are three small organizations that looked at the process and measured the pain before they bought anything, as reported by an independent body, not a vendor. The figures are theirs.

CPA Australia

Smart Business Solutions Group

A 22-person accounting, mortgage broking and financial planning firm. Staff shortages pushed routine work offshore; time zones and quality brought it back. They named two jobs: Business Activity Statements and the sensitive paperwork they could not outsource.

The first fix was wrong. A consultant charged $3,200 per automated process and $5,000 a month in licensing, and the bot broke with every software update. The gap was between the accountants who knew the process and the programmers who did not.

The second fix was theirs. Bots now process BAS and Instalment Activity Statements and prepare work papers, and AI drafting aids are used only with employee review and sign-off. In FY2025 that contributed to a 14% revenue uplift and a 23.6% reduction in administrative wages, with two long-term roles replaced by one junior through natural attrition, not layoffs.

Source: CPA Australia, "How one accounting firm implemented AI"

NAB (National Australia Bank)

Bella Manufacturing

A custom manufacturer of food trucks, emergency-services kitchens and community-service vehicles. The owner's administrative burden, including reviewing complex documents and comparing insurance quotes, was taking time away from the factory and clients.

The measurement was a side-by-side test. Start with one small task, give it to the AI tool, do it yourself, and compare the result and the time. The first test was an insurance quote comparison.

The result: several thousand dollars saved, hours of work returned, and a potentially under-insured area found and fixed. The owner keeps human oversight on every output, and says it stays essential in bespoke manufacturing.

Source: NAB, "How are small businesses using AI"

America's Credit Unions

FORUM Credit Union

Loan processing was the bottleneck. Classifying documents and preparing files for underwriting took staff time that capped how many loans the team could move.

The measurement was minutes per file. Document classification took 15 to 20 minutes. Underwriting preparation took 45. Those two numbers defined the problem and set the target.

The result: classification fell to seconds with accuracy above 95%, underwriting preparation fell from 45 minutes to five, and loan volume grew 70% without adding staff. The stated payoff was doing more with the same people.

Source: America's Credit Unions, "Credit unions deliver exceptional member experiences through AI"

Your results will depend on your process and your baseline. These accounts are here because each one followed the order above: map, number, fix, review.

What We Do

Service areas

Every engagement is scoped to the problem, not a rate sheet. These are the kinds of work that come out of step three.

📊

Process Mapping & Measurement

How your work actually moves, where the pain points are, what each costs today in hours, errors, or delays, and the baseline we will measure against. For businesses that know something is wrong but cannot yet say what. Often this is the whole first engagement.

⚙️

Workflow Automation

Redesign the hand-offs that fail, then automate the repetitive steps, often in Microsoft 365 and Power Automate, with the compliance requirements built into the flow instead of bolted on afterward.

🤖

AI Readiness & Implementation

Put AI on the first pass of the work where the measurement says it fits: summarizing, drafting, comparing, classifying. With written rules for what it may see and a named person reviewing every output before it leaves the building.

🔧

Technology Stack Rationalization

Inventory the tools you pay for, retire the overlap, and connect what remains so the stack works as one system. Usually funded out of the subscriptions it cancels.

☁️

M365 & Integration Advisory

Microsoft 365, Entra ID, and Power Automate expertise, and the integration work to connect business and compliance tools. If you run on M365, you almost certainly already own capability you are not using.

🖥️

For MSPs: Operations & AI Inside Your Practice

The same method applied to the MSP itself: service desk triage, onboarding and offboarding, patch orchestration, reporting, and AI adoption, from someone who has run an MSP services division. Details on the MSP partnership and consulting page.

Not sure which process to fix first?

Read how to choose the first process to fix, a method you can run yourself in a week. Or start with a conversation about the one that is eating yours.

Schedule a Call