Choose the AI consultant who can show you something already working, puts a fixed price in writing before you commit, and can tell you exactly what happens when it breaks. The twelve questions below test for exactly that.
We're writing this from the other side of the table. Webiske builds AI automation in Louisville for small and mid-size businesses, so we sit through a lot of first calls. These are the questions we'd want answered if we were the ones hiring.
What does an AI consultant actually do for a small business?
A useful one finds the repetitive work that eats your team's hours, builds something that handles it inside the tools you already use, and keeps it running. For most small businesses that means phone answering, review replies, follow-up emails, website chat and simple reports. It rarely means a company-wide AI strategy.
The title covers very different businesses. Some consultants write a report and leave, some resell one chatbot to everyone, and some build and manage custom systems. A report is fine if you have someone in-house to build what it recommends. If you don't, you need someone who builds it and stays.
How does AI consulting work for a company?
Most engagements follow four steps: a short call to see if automation fits, an assessment of one workflow to find where the time goes, a build that plugs into your existing tools, and ongoing support. Be wary of anyone who skips the assessment and goes straight to a quote.
- The call. You describe the work that's jamming up. They tell you whether it's a fit. An honest "no" at this stage is a good sign.
- The assessment. Someone watches how the work really happens: who types what, where, and how often. You should come away with a written scope and a fixed price.
- The build. The automation gets wired into your email, CRM, calendar or phone. You should see it handle real examples before it goes live.
- Ongoing support. Someone monitors it, fixes it when one of your tools updates, and adjusts it as your business changes.
12 questions to ask an AI consulting provider
Ask about ownership, failure, human approval, integrations, measurement and references before you ask about features. Features are easy to demo. What happens in month four, when a tool updates or a customer gets a strange reply, tells you far more.
- What exactly will this do, in one sentence? If they can't say it plainly, the scope isn't settled yet.
- Can I see it running, ideally on your own business? A phone line you can call or a tool you can try is worth more than a slide deck.
- Who owns the accounts and the data? Phone numbers, software logins, AI service accounts and your customer data should be in your name, or handed over on request. Get it in writing.
- What happens if we stop working together? You should be able to keep what you paid for, or at least get your data and settings back cleanly.
- Where does a person approve things before a customer sees them? Anything customer-facing, especially complaints and bad reviews, should wait for a human until you decide otherwise.
- What happens when it breaks, and who finds out first? The answer you want is that their monitoring catches it before your customers do, and you know who to call.
- Which of my tools does it plug into? Name your email, CRM, calendar, phone system and booking software. Ask whether your team will need a new login or app to check.
- Where does my data go? Ask which outside AI services see your information and what they're allowed to keep. If you handle patient or legal files, ask this one first.
- How will we know it worked? Agree on a number before the build, like missed calls per week or hours spent on follow-up. Measure it before and after.
- What would you tell me not to automate? A consultant with no answer here probably hasn't looked closely at your work.
- Is the price fixed before I commit, and what's ongoing? You want the build price and the monthly cost in writing before work starts.
- Can I talk to someone you've done this for? Some firms, us included, don't name clients. If so, ask to see the system working instead.
How should I read reviews and ratings for AI consulting firms in Louisville?
Use them, but don't stop there. AI automation is new, so a local firm's reviews often describe older work like websites or marketing. Look for reviews that describe a specific system, then confirm it yourself with a demo and a reference.
A five-star review for a logo doesn't tell you whether their phone bot handles a confused caller well. Also look at how the firm answers its own reviews. Slow or canned replies are a preview of how they'd treat your customers.
Red flags to watch for
Walk away from anyone who won't put a price in writing, keeps your accounts in their own name, can't show anything running, or says the AI never needs a person. Promised results with no plan to measure them is another one.
- A long contract before you've seen it work on your own business.
- No human approval step on messages that reach customers.
- Vague answers about where your data goes.
- Nobody named to call when it breaks.
What drives the cost of AI consulting for a small company?
Scope drives cost. The number of workflows, how many tools the automation has to connect to, how messy your data is, and how much monitoring it needs afterward all move the price. A firm worth hiring looks at your work first and gives you a fixed number before you commit.
- Number of workflows. One job, like review replies, costs less to build and run than five connected ones.
- Integrations. Some tools connect easily. Older or locked-down systems take more work, and sometimes a workaround.
- Data cleanup. If your CRM is full of duplicates and half-filled records, someone has to fix that before automation can rely on it.
- Voice or text. Answering the phone has more moving parts than drafting emails.
- Ongoing monitoring. Tools update and businesses change. Someone has to watch it and fix it, which is why this work usually carries a monthly fee alongside the build.
How can a small business safely start using AI tools?
Start with one repetitive job. Keep a person approving anything customers see, keep every account in your own name, and don't paste customer or patient details into free AI tools. Measure the job before you change it so you can tell whether it helped.
Good first jobs are frequent and low-risk: answering routine calls, replying to good reviews, drafting follow-up emails for someone to send. Bad first jobs involve money, legal wording, medical information or angry customers. Once the first job has run quietly for a while, add the next one.
How Webiske works
Webiske is a Louisville company that runs its own operation on AI and builds the same kind of systems for other businesses. Every engagement starts with a free 30-minute feasibility call: a straight answer on whether automation fits your business, before you pay anything.
When it does fit, the next step is an operational assessment at a fixed fee, agreed up front, for one workflow. We shadow how that work really happens and come back with the arithmetic behind every opportunity we find, including where the honest answer is that a step shouldn't be automated at all. The fee is credited in full toward your build if you contract the build within 90 days of delivery. Then we build it into the tools you already use and manage it on a flat monthly retainer.