Ask a founder why they have not built AI into their business yet, and the honest answer is rarely the price. It is the time. They picture another sprawling project, another thing that needs a project manager, another set of meetings their team cannot absorb. So it stays on the someday list.
Here is the number that should change that calculation. A done-for-you AI Build takes about four hours a week from one person on your team. Not four hours from everyone. Four hours from a single designated point person, across an eight to twelve week engagement. That is the real time cost of installing a working AI employee.
If you have been treating AI as a project you cannot staff, this is the post that shows you exactly where those four hours go, and why the rest of the work does not land on your desk.
How much time does an AI implementation actually take?
For a done-for-you Build, plan on about four hours per week from one point person, plus a 90-minute kickoff at the start and three 60-minute reviews spread across the engagement. Over an eight to twelve week build, that is roughly 35 to 55 total hours from your side, concentrated in one person rather than scattered across your team.
That is the whole footprint. No daily standups, no all-hands, no parallel workstream that pulls three people off their real jobs.
Where the four hours actually go
The four hours are not busywork. They are the specific inputs only your team can provide, the things no outside firm can invent for you.
- Access and context. Early on, your point person gets us into the tools and shares how the work really happens: the SOPs, the voice, the edge cases, the unwritten rules. This front-loads most of the time.
- Decisions, not production. Through the build, the weekly time is spent answering questions and making calls. Which refund cases get escalated. What the brand would never say. Which step a human always keeps. You decide; we build.
- Review and correction. At each review, you see the employee working on real inputs and tell us where it is off. Those corrections are what sharpen it. This is the highest-leverage hour you will spend, because it is what makes the output something you would sign off on without checking.
Notice what is not on that list. You are not writing prompts, wiring tools, testing edge cases, or documenting anything. That work exists, it is just on our side.
Why most of the work stays off your plate
The reason the number is four hours and not forty is structural. We are not handing your team a tool and a login and wishing them luck. We are installing an employee and doing the building ourselves.
A real AI employee has four layers: a brain trained on your business, the skills to do a specific job, the tools to act inside your systems, and a memory that improves with use. Constructing those layers, training them, connecting them, and testing them is the labor. It is skilled work, and it is exactly the work that stalls when a busy internal team tries to do it between their day jobs.
The model is the part everyone can do. The other three layers are where the hours live, and where projects quietly become shelfware when no one has time to finish them.
The math that makes four hours worth it
Four hours a week for a couple of months is the input. The output is what makes it lopsided in your favor.
In one of our builds, the AI support employees took 15 hours a week of support work off the team. Permanently. That is recurring time back, every single week, on work people used to do by hand.
Run the comparison honestly. You spend roughly four hours a week for the length of the build, once. In return you get hours back every week for as long as the employee runs, and you own it outright. The investment ends. The return does not.
| Your four hours a week | What it buys |
|---|---|
| One point person, not the whole team | The rest of your people stay on their real work |
| Decisions and context, not production | We do the building, training, and testing |
| A few weeks, once | Hours back every week, ongoing |
| Reviews that sharpen the output | An employee you trust without checking |
The hidden cost of doing it yourself
The alternative to four guided hours a week is not zero hours. It is far more hours, spread thin, with a much higher chance of nothing shipping.
MIT’s 2025 State of AI in Business report found that AI systems delivered by outside specialists succeeded roughly twice as often as internal builds. The reason is rarely talent. It is time and focus. Internal AI projects compete with everything else on a busy team’s plate, so they get built halfway, lose momentum, and stall before they reach production.
Doing it yourself feels cheaper because the hours are hidden inside existing salaries. They are not free. They are just uncounted, and they usually add up to more than the engagement would have asked of you, with a worse result.
What you should ask before any AI engagement
If a firm cannot tell you exactly how much of your team’s time they need, that is the answer. Before you commit, ask:
- How many hours per week, and from how many people?
- What specifically do you need from us, and when?
- What happens on your side versus ours?
- What do we own when it is done?
A clear, honest time commitment is a sign the firm has actually done this before. A vague one is a sign you are about to become the project manager.
Four hours a week is a decision, not a burden
The time objection is real, but it is pointed at the wrong target. You are not signing up to run an AI project. You are signing up to spend a few focused hours a week making decisions only you can make, while someone else does the build. The work that scares you off is the work we take.
That is the whole point of a done-for-you engagement. You stay on the business. We install the workforce. And the hours you put in come back to you, every week, long after the build is done.
Frequently asked questions
How much time does an AI implementation take from my team? For a done-for-you Build, about four hours per week from one designated point person, plus a 90-minute kickoff and three 60-minute reviews across an eight to twelve week engagement. That is roughly 35 to 55 hours total, concentrated in one person rather than spread across the team.
Who on my team needs to be involved? One point person carries the weekly time. Others join briefly for short interviews about how their part of the business works, but no one else takes on an ongoing commitment.
What do those four hours actually involve? Giving access and context early on, answering decisions through the build, and reviewing the employee on real work so corrections can sharpen it. You make the calls. The building, training, and testing happen on our side.
Why does it take so little of my time? Because it is done for you. Constructing the four layers of an AI employee, the brain, skills, tools, and memory, is the real labor, and we do it. Your time goes only to the inputs and decisions no outside firm can provide.
Isn’t it cheaper to have my own team build it? Usually not. MIT’s 2025 research found vendor-built AI systems succeeded about twice as often as internal ones. Internal projects compete with everyone’s day job, so they stall. The hidden hours of a DIY build often exceed a guided engagement, with a lower chance of shipping.
What do I actually get back for the time I put in? An AI employee you own and a recurring return on hours. In one of our builds, AI support employees took 15 hours a week off the team, every week, from work they used to do by hand.
How is the time structured across the engagement? It front-loads. The kickoff and early access sessions take the most time, then the weekly load settles into short decision and review cycles. By the handoff, your team is spending time managing the employee, not building it.
Curious what four hours a week could install in your business? Book a strategy call, or start with a Map and get a ranked plan in three weeks.
Lisa Hayford is the founder of The Lasting Method, where she designs and installs AI employees for DTC brands doing $1M to $10M in revenue. She has built and rolled out full AI implementations across entire operations, from marketing and support to operations and finance, with one rule: the business owns what gets built, not a vendor. She writes about what it actually takes to make AI stick inside a company instead of becoming shelfware.