At some point every operator who takes AI seriously asks the same question: do we hire someone, or do we bring someone in? The question usually gets framed as cost, and cost matters. But the two options also see different things, and that difference decides more of the outcome than the invoice does.
I've worked on both sides of this. I'm VP of Technology at Summit Sotheby's International Realty, and through Elorati I'm the outside studio for other operators. What follows is what each seat actually gives you, and how to tell which one your business needs right now.
What an in-house person learns that nobody can hand them
Every business has a theory of how its work gets done. It lives in the org chart, the process docs, and the way leadership describes the week. Then there is what the work actually is: where it really slows down, who quietly holds three jobs together, which step everyone skips because it never made sense, and how the team and the work push on each other.
An in-house person learns the second version. Nobody briefs them on it. They pick it up by sitting in the middle of it for months, watching which requests come back twice, which handoffs drop, and which "quick fix" turns into a standing task. That knowledge is the most valuable thing an AI project can start with, because automation built on the theory of the work automates the wrong thing. It speeds up a step that was never the bottleneck and leaves the real one alone.
An outside firm can interview its way toward that picture. It will still miss things, because the parts that matter most are the parts nobody thinks to say out loud.
What an outsider can see that an insider can't
The inside view has a cost: history. Every process in a business was built by someone, and the people who built it are usually still around, still respected, and still attached to how it works.
Picture the spreadsheet a well-liked ops manager built years ago. Everyone works around it, half the team keeps a side copy, and nobody wants to be the person who suggests retiring it, because retiring it feels like a verdict on someone they like. An insider knows exactly what's wrong with that spreadsheet and has every reason to leave it alone.
An outside advisor doesn't carry that weight. They can look at the process without the relationships attached, call the spreadsheet what it is, and propose the replacement without it reading as a judgment on a colleague. They also bring range. Someone who has seen the same problem in a dozen businesses knows which fixes hold up and which ones only demo well.
The deciding question: steady work or bursts?
Set the two kinds of vision aside for a moment and the money question is fairly simple. Hire a person when you have ongoing projects and keeping someone in-house costs less than paying a vendor for the same work. Bring in an advisor when the work comes in bursts.
Steady work looks like a backlog that never empties: systems to maintain, a queue of internal requests, new tools to roll out and train people on every quarter. If you can picture a full week of real work for that person fifty weeks a year, the math usually favors a hire, and you get the inside view as a bonus.
Burst work looks like a decision followed by a quiet stretch: choosing between vendors, scoping a first build, reviewing a pilot before you scale it, untangling a project that went sideways. Paying a full salary to cover a few intense months a year means paying for a lot of idle capacity, or inventing work to fill it.
A rough test before you decide: list the AI and software work you expect over the next twelve months, and mark each item as recurring or one-time. If the recurring column fills a job, hire. If most of the list is one-time decisions, start with an advisor.
When to hire in-house
Hire a person when:
- The work is ongoing. There's a steady stream of systems to run, requests to handle, and tools to keep current, and it adds up to a real job.
- The inside view is the hard part. Your bottlenecks live in team dynamics and undocumented habits that only someone present every day will find.
- You can give them room. An in-house hire buried in help-desk tickets never gets to the work you hired them for. Protect the time or the hire won't pay off.
If all three are true, an outside firm is the more expensive way to get a worse result, and that includes us. The honest advice in that situation is to hire well and bring in outside help only for the occasional second opinion.
When to bring in an advisor
Bring someone in from outside when:
- You're early. Before the first budget gets spent is the cheapest time to get an outside read. Choosing the wrong vendor or scoping the wrong build costs far more than the advice would have.
- The work comes in bursts. A vendor decision, a pilot review, a first build. Intense for a few weeks, then quiet.
- You need a cold read. The process everyone is protective of, the tool nobody wants to admit isn't working, the project that has been "almost done" for a quarter. Someone without the history can say it plainly.
- You don't know yet what the job is. It's hard to write a good job description for a role you can't define. An advisor can help you figure out whether you need a hire at all, and what that person should actually do.
The setup most operators end up with
In practice the answer is often both, at different times. An outside advisor helps make the early decisions and scope the first builds. As the work turns from decisions into upkeep, it either moves to an in-house owner or stays with the people who built it on a managed arrangement. Later, the in-house person brings the outside view back in for the big decisions, when a fresh set of eyes is worth more than another month of context.
What matters is that someone owns the work at each stage. The worst version is the one where nobody decides: no hire, no advisor, and a team trying to fit AI into the gaps of their actual jobs. That version produces a lot of trials and very little that sticks.
Frequently Asked Questions
Is it worth hiring an AI consultant instead of an employee?
It depends on whether the work is steady or comes in bursts. If you have enough ongoing AI and software work to fill a job, an employee usually costs less over a year and builds an inside understanding of your business that an outsider can't match. If the work is mostly one-time decisions, like choosing a vendor, scoping a first build, or reviewing a pilot, a consultant is cheaper because you pay for the intense stretch and nothing else.
Should I hire an in-house AI team or outsource AI strategy?
Most operators at small and mid-size companies don't need a team, they need either one capable person or occasional outside help. Start by listing the AI work you expect over the next year and marking each item as recurring or one-time. Recurring work that adds up to a job points to a hire. A list of mostly one-time decisions points to an advisor.
What does an AI consultant actually do for a small business?
A good one helps you make the expensive decisions before you make them: which problems are worth automating, whether to build or buy, which vendor to trust, and how to scope a first project so it survives contact with real work. They also give you a read on your processes from someone who isn't attached to how things have always been done.
Can I start with a consultant and hire someone later?
Yes, and for many operators it's the better order. An advisor can help you work out what the in-house role should be, what it should own, and what a good candidate looks like, which is hard to do before you've seen the work up close. When the recurring work grows large enough to fill a job, that's the signal to hire.
This guide provides educational information based on industry research and case studies. Individual results vary by market, budget, and execution.