The short answer
Start with a consulting firm and name someone inside the business to own the work. Hire a full-time AI person later, once the work keeps coming. A firm can usually ship a first automation in weeks, while a new hire needs time to be recruited and then to learn your business before shipping anything. When you have more AI work than an outside team can handle part time, a full-time person becomes worth the salary.
That's the rule of thumb we share with clients at PerezCarreno & Coindreau, and it comes from 18+ years of building software for businesses, most recently teams with 5 to 200 employees. Some businesses should hire first, though, and the signs are easy to spot once you know them.
What an in-house AI person does and what a consulting firm does
An in-house AI person is an employee whose job is to build and maintain AI tools for your business. A consulting firm is an outside team you hire for a defined project with a set scope and price. The in-house person stays with you day after day. The firm brings experience from many businesses and steps back once the project and its support period end.
Small businesses use "in-house AI person" for two quite different jobs, so it helps to be clear about which one you mean:
- An AI engineer or developer who writes code, connects systems, and builds custom agents. This is a technical hire and the harder one to recruit.
- An AI champion or operations lead who already knows your business, learns to use AI tools well, runs simple automations, and works with outside builders. This is often an existing employee who gets training.
Most small businesses need the second role long before they need the first. A champion can learn in a hands-on workshop, get ongoing support from an advisor, and become the bridge between your team and whoever builds the bigger systems.
Consulting firm vs in-house hire, side by side
Neither option wins everywhere. The firm wins on speed and flexibility at the start, and the in-house person wins on day-to-day availability and on knowing your business better every month. The table shows where each one is stronger.
| Point | Consulting firm | In-house AI hire |
|---|---|---|
| Time to a first working result | Weeks. Our first automation is usually live within two weeks | Months, after recruiting and time to learn your business |
| Cost shape | A fixed project fee that ends when the work ends | Ongoing salary, benefits, tools, and management time |
| Range of skills | A team that has seen many businesses and tools | One person's strengths and gaps |
| Knowledge of your business | Learned during discovery, then handed over in writing | Grows every month the person stays |
| Availability | Scheduled project time and a support window | Every workday |
| If it turns out to be the wrong fit | You end the project | A hard hiring decision to undo |
| Best fit | The first 3 to 5 automations, or work you can't define yet | A steady stream of AI work, usually 10+ workflows |
What an in-house AI hire costs beyond salary
The salary is only one part of what an in-house AI hire costs. You also pay for benefits and payroll taxes, recruiting, and software and AI subscriptions. Then there are the months a new person needs to learn your systems before they ship anything useful. Add the time a manager spends guiding someone in a role nobody else fully understands.
We don't publish salary figures here because they change quickly and vary a lot by city. Look up current ranges for your area before you budget, then add the costs above. Compare that yearly total with a fixed project quote for the same first year of work. At PerezCarreno & Coindreau, most first projects cost $3,900 to $11,900.
There's also a skills risk. One person can't be strong at everything AI work needs, from planning and data cleanup to integrations, testing, and teaching coworkers. If your first hire is great with prompts but weak at connecting systems, you'll find out months in.
A firm spreads that risk across people who have done the work many times. PerezCarreno & Coindreau has built AI and automation for 100+ businesses in 11 industries, so we've usually seen a version of your problem before.
When an in-house AI hire is the better choice
Hire in-house when AI work is continuous and central to how you operate. If you can point to a backlog of automations that would keep a person busy all year, or AI is part of the product you sell, a full-time hire is usually worth it. You also want someone inside when your systems change every week and waiting on an outside team would slow you down.
Signs it's time to hire:
- You already have 10 or more workflows built or planned, and they need regular care
- AI is part of the product or service you sell to customers
- Your systems change often, and outside turnaround times would slow the business
- You have strict rules about who can touch your data, and outside access is hard to approve
- You have a technical leader who can manage and mentor the hire
If most of those fit, hire. A consulting firm can still help by building the first systems the new hire will own and by coaching them through their first months.
When a consulting firm is the better choice
A consulting firm is the better choice when you have a handful of clear problems and no proof yet that AI work would fill a full-time job. A fixed-scope project gets a working system live in weeks, and the spending stops when the project ends. You learn how much AI work your business has before you commit to a salary.
Signs a firm fits better right now:
- You have a few workflows you'd like automated and no AI roadmap beyond them
- You don't have a technical person who could manage an AI hire
- You need results this quarter and can't wait out a hiring cycle
- You aren't sure which skills the role would need
- You want a fixed, written price before you commit
That describes most of our clients. They're businesses with 5 to 200 employees whose teams lose 10 to 20 hours a week to admin work that can be automated. For them, a Custom AI Automation project usually takes 2 to 4 weeks, and the team owns the system from day one.
The hybrid path most small businesses take
The path that works for most small businesses is a sequence. A consulting firm plans and builds the first systems while an internal owner learns to run them, and you hire technical staff only after the work proves it needs a full-time person. Each step lowers the risk of the next one.
- Start with an audit. An AI Strategy & Audit takes 1 to 2 weeks and ranks the 3 to 5 automations worth building first. The report is vendor-neutral, so you can hand it to any build team, including your own engineers.
- Build the first one or two with outside help. Your first automation is usually live within two weeks.
- Train an internal owner. A half-day or full-day workshop gets the team using AI on its own work, and the owner learns the system during the handoff.
- Keep an advisor on call. AI Coaching for Business Owners gives you two calls a month plus Slack access, billed month to month.
- Hire when the backlog proves it. Once you're past roughly 10 workflows and still adding more, a full-time hire has plenty to do from the first day.
The math works in your favor along the way. The groundwork from your first project makes every later project faster, so by the third or fourth workflow, new automation costs much less. If you do hire, your new person inherits working systems with written instructions instead of a blank page.
How to keep the knowledge in-house when you use a firm
The fair worry about any outside firm is that the knowledge leaves with them. You avoid that by making the handoff part of the scope. Put ownership of every account and written instructions into the agreement before the project starts, and ask for training for the people who'll run the system day to day.
At PerezCarreno & Coindreau, every build includes:
- Accounts, credentials, and code in your company's name
- A runbook that explains how the system works and what to do when it fails
- A hands-on training session and a recorded walkthrough
- Error alerts sent to someone on your team
- 30 days of support after launch on automation builds, and 90 days on custom development
One client review describes the goal well. Melody Meilleur-Durand of Bia Education wrote that Armando "always makes sure to empower his clients to be as self-sufficient as possible."
Common mistakes on both paths
The most common mistake is hiring a full-time AI person before anyone has written down the work they'll do. The second is using an outside firm without planning the handoff, so the knowledge leaves when the project ends. Both are easy to avoid if you describe the work first and make ownership part of every agreement.
Other mistakes we see small businesses make:
- Hiring a developer when the real need is a trained operations person
- Writing the job description from a vendor's sales pitch instead of your own workflows
- Letting an outside firm keep the accounts or code in its own name
- Buying AI tools for everyone without training anyone to use them
- Skipping measurement, so nobody can say how much time the first project saved
Buying tools without training is one of the failure patterns we describe in our AI adoption guide, and it's one of the cheapest to fix. A half-day workshop gets people using AI on their own tasks, and it often surfaces the next workflow worth automating.
Where PerezCarreno & Coindreau fits
PerezCarreno & Coindreau works as the outside AI team for small businesses that aren't ready for a full-time hire, and as a partner for those that are. Founder Armando J. Perez-Carreno leads every project and writes the production code himself. We work in English and Spanish, on-site around San Antonio, Texas, or remotely anywhere in the United States.
If you're weighing a hire, a free 30-minute call is a good place to test the idea. We'll look at your work with you and tell you honestly whether it points to a project or a hire. You can also read how outside help compares in our guides on consultants and freelancers and on alternatives to big consulting firms.