How to Implement AI in Your Business: The 90-Day Roadmap We Build in Every AI Opportunity Audit

AI IMPLEMENTATION

By Evan Dorian, founder of Dorian Media Group (established 2012). Last updated November 2026.

To implement AI in a small business, pick one repeatable process that costs your team hours every week, measure how long it takes today, simplify it, then automate what is left with tools you already pay for. Run that one system for 30 days and confirm the hours saved before you build the next. The whole cycle fits in 90 days.

This is the order we follow at Dorian Media Group. We have built brands, websites, photography, video and marketing systems for clients since 2012, and our AI consulting practice grew out of that work: the proposals, follow-up emails and weekly reports we used to assemble by hand are now the first things we automate. Below is the 90-day roadmap we sequence in the AI Opportunity Audit, the tools we build in, our prices, and how to tell whether it paid off.

Where should a small business start with AI?

Start with work that repeats every week, follows rules someone could write down, and already lives in digital tools. Proposals, lead follow-up, weekly reports, support replies and copying data between systems usually qualify. Leave judgment calls, pricing decisions and anything built on messy data for later.

When we sort a list of candidate processes, five questions decide the order:

1. Does it happen at least weekly? A task that runs once a quarter rarely pays back the build.
2. Could you write the rules on one page? If the steps change every time, the AI will guess, and you will spend the saved time checking its guesses.
3. Are the inputs already digital? Paper files and the owner’s memory cannot be automated yet.
4. Is a mistake cheap to catch? A draft reply that a person approves is low risk. An invoice that sends itself is a different conversation.
5. Does someone own it? Every system needs one named person who notices when it stops working.

A process that earns five yeses is a first build. Three or four usually means a short cleanup first: write the steps down, merge the duplicate spreadsheet, agree who owns it. If you want a structured way to check this, score yourself with our 20-question AI readiness scorecard.

The 90-day AI implementation roadmap

A 90-day AI roadmap has four stages: an audit that ranks your opportunities (week 1), a small first win (week 2), one or two working systems measured against a baseline (weeks 3 to 10), and a review that plans the next quarter (weeks 11 to 13). Each stage has a gate you pass before moving on.

Weeks 1 and 2: the audit and a first win

The audit starts as a conversation. We ask where your time actually went yesterday, what you dread but cannot hand off, and what would change your week if it ran itself. The report includes a priority matrix that says what to start first, what comes next and what to skip, plus a financial breakdown of what each recommendation costs and returns. The first win is deliberately small: it shows whether the team will use AI before anyone pays for a build.

Weeks 3 to 10: build one system, then measure it

Build one system at a time. A workflow automation build with us typically takes 1 to 2 weeks, and a marketing automation or knowledge base build 2 to 3 weeks, so the build fits inside this window with a full month left to measure it.

That month matters more than the build, because a system that looks great on launch day can be abandoned by week three. If it saves what it promised, the next process on the ranked list can start around week 7. If it falls short, fix it before adding anything. The review in weeks 11 to 13 compares confirmed hours with the audit estimate and sets the next quarter.

Which AI tools does a small business actually need?

Most small businesses need three kinds of tools: an AI assistant such as Claude or ChatGPT for drafting and summarizing, an automation platform such as Zapier, Make or n8n to move information between apps, and the systems already running the business, like your CRM, email platform and Google Workspace. We build inside that existing stack.

If a build genuinely needs a new tool, the written scope says so, with the monthly cost attached. Subscription prices vary by plan, so the audit lists them per recommendation.

What does AI implementation cost with Dorian Media Group?

The AI Opportunity Audit is $997 one time. AI Concierge, where we build with you, is $1,750 a month for two 45-minute sessions, or $2,950 for four. Fixed-price AI Builds run from $1,000 to $5,000 depending on type, and larger engagements start at $5,000.

Join Concierge within 30 days of your audit and the $997 is credited to your first month. The credit does not apply to builds. Every build includes 30 days of fixes, and you own every prompt, workflow and document. To compare these prices with hourly and retainer models, read what AI consulting costs. Prices are the same whether you work with us as an AI consultant in Kansas City or as an AI consultant in Newport Beach and Orange County.

How do you measure the ROI of AI?

Measure AI ROI in hours first. Time the process before you change it, time it again after the system has run for 30 days, and have the person who does the work confirm the difference. Multiply the confirmed hours by what that time costs, then subtract tool and build costs. Estimates do not count until someone confirms them.

1. Baseline. For two weeks, log how long the process takes and how often it runs.
2. Run. Put the system into normal use for 30 days and keep the old method available for exceptions.
3. Confirm. The person who owns the task states the new time. In AI Concierge this goes into the Build Log, and anything not yet confirmed stays marked pending.
4. Price it. Monthly hours saved, times the loaded hourly cost of whoever did the work, minus the monthly tool cost. Divide the build cost by that figure to get months to payback.

A worked example with made-up round numbers, not a client result: a task that took 6 hours a month now takes 2. Four hours at a $50 loaded hourly cost is $200 a month. Subtract $30 a month in tools and the net is $170, so a $1,000 build pays for itself in about six months.

Faster lead replies and fewer missed follow-ups matter too, but they are harder to attribute. Track them separately and keep them out of the hours figure.

Does a small business need an AI policy?

Yes, and one page is enough to start. A small business AI policy names the approved tools, lists the information that never goes into them, requires a person to review anything customer-facing or financial, and names an owner for each automated system. Write it in week one and revisit it at the 90-day review.

• Approved tools. List the tools staff may use for work, on business accounts the company controls rather than personal logins.
• Information that stays out. Decide what never goes into an AI tool, such as confidential client files, payment details, passwords, and health or personnel records. Check each tool’s admin settings for how it stores and uses your data.
• Human review. Anything sent to a customer, anything that quotes a price and anything that commits the business gets approved by a person first.
• Ownership. Every automation has a named owner and a written how-to-run-it document.
• Disclosure. Decide when you tell customers that AI helped draft something, and apply the rule consistently.

For a fuller framework, the NIST AI Risk Management Framework is free and written for organizations of any size. This is practical guidance, not legal advice. If your industry is regulated, check your obligations with your own counsel.

AI implementation FAQ