Most owners can describe how a lead becomes a scheduled job. But the version in the owner's head is often not the version the office runs on a busy Tuesday. Somewhere between the call and the crew, people have built workarounds to get the job done.
Employee workflow interviews are how you find that gap before you build AI support around the wrong process. This post is for owners and operations managers at service businesses with roughly 5 to 49 employees who are considering a first AI employee or AI Team and want the design to match how the work actually happens.
The Business Problem
When an owner describes a workflow, they usually describe the intended one. When a dispatcher, estimator, or office coordinator describes it, you hear something different:
- "I copy the notes from the CRM into a text because the techs never open the app."
- "If the customer calls back, I check three places to see who talked to them last."
- "I keep a spreadsheet of estimates that need a second follow-up because the system only sends one."
- "I wait for the owner to approve anything over a certain amount, but that rule is not written down."
These are not failures. They are people filling gaps the system does not cover, and they are exactly the details an AI employee needs to know. If an AI Task Expert is built around the owner's version of the workflow, it will miss the workaround, duplicate it, or break it.
What This Costs the Business
Skipping the people who do the work usually shows up later as rework:
- Wrong inputs. The AI employee is told to read information from a field that nobody fills in.
- Missed handoffs. The real handoff happens by text or hallway conversation, so the AI never sees it.
- Quiet resistance. Employees who were not asked tend to see the new support as something done to them, not with them.
- Owner dependency stays in place. If the hidden approval rules are not surfaced, everything still routes back to the owner.
The owner ends up carrying the same mental load as before, plus a new system that does not quite fit.
What Should Happen First
Before anyone talks about AI, interview the people closest to the work in question. Keep the scope narrow. If the first build is estimate follow-up, talk to the people who create, send, and chase estimates. You do not need to interview the whole company to design one Task Expert.
A good interview is a conversation, not an audit. Ask people to walk through a real recent example. Useful questions include:
- What starts this work for you, and how do you find out about it?
- Where do you go to get the information you need?
- What do you do when that information is missing or wrong?
- Who do you hand the work to, and how do they know it is ready?
- What do you track outside the main system, and why?
- What decisions do you wait on, and who makes them?
- If you could stop doing one part of this, which part would it be?
The last question matters. It often points straight to the first AI employee worth building.
How to Keep the Answers Honest
Employees will not describe workarounds if they think they will be blamed for them, or if they think the interview is about replacing them. A few practices help:
- Explain the purpose up front. The goal is to remove repetitive work and missing information, not to grade anyone.
- Interview without the owner in the room when possible. People speak more plainly to a neutral party.
- Treat workarounds as useful information. A workaround is evidence of a real need the system does not meet.
- Share what you heard. Summarize the findings back to the team so they can correct anything that was misunderstood.
This also starts adoption early. People who helped shape the design are more likely to use it. Our guide on how to introduce AI to your employees covers what comes after the build.
Where AI or Systems Can Help
Interviews turn into design decisions. Each finding should answer a practical question:
- For a Task Expert: What exactly triggers the job, what information does it receive, what does it produce, and who reviews it? A clear job description for an AI employee comes straight from these answers.
- For human employees: Which judgment calls stay with people, and which approval rules need to be written down?
- For an AI Team: Where do handoffs between Task Experts happen, and who manages the team? A manager or director role inspects outputs and escalates exceptions to a named person.
- For your existing systems: Is the CRM missing a field, a status, or a notification that people are covering by hand?
On the system question, interviews usually land in one of five places. No change is needed because the workaround is a habit, not a system gap. A light configuration change, such as a new field or status, removes it. An integration is needed because information lives in two tools that do not talk. A new CRM backbone is worth discussing because the current system cannot support the workflow at all. Or the answer cannot be determined yet and needs more discovery. In most cases the existing CRM stays.
What Changes and What Stays the Same
What changes: hidden steps get written down, workarounds get replaced or supported, and the first AI employee is designed around the real workflow.
What stays the same: your team, your customers, and most of your tools. Interviews often confirm that much of the current process works, which tells you what not to rebuild.
When On-Site Discovery Helps
For a single, bounded job in a well-documented business, a few remote interviews may be enough. Our post on choosing between a remote AI build or an on-site assessment explains how to tell.
For builds that touch several departments, a CRM configuration, or executive support, interviews work better in person. StrategixAI may spend one, three, or five days on-site, sitting with the people doing the work and watching it happen, so the AI employees, handoffs, and reports are designed with minimal ambiguity. Observation catches things interviews miss, like the sticky note on the monitor or the second phone in the truck.
Where StrategixAI Fits
StrategixAI helps service businesses understand how work actually moves before designing AI support. That includes leadership and employee interviews, workflow mapping, system review, and a clear plan for the first AI employee or AI Team. As the business changes, the people closest to the work keep surfacing new needs, which is why AI Teams benefit from ongoing maintenance rather than a one-time setup. You can see how an engagement is structured on our how we work page.
Practical Next Step
Pick one workflow that frustrates you. Ask two people who do that work to walk you through their last real example, and write down every step, tool, and workaround they mention. Compare it with how you would have described it.
If the gap is bigger than expected, discuss whether an on-site business assessment would help clarify your first AI Team build.