AI without an AI hire: five ways to get the work done, and what each leaves you.

Five ways to get AI work done.

Getting AI working in a business takes five kinds of work: finding the workflows worth doing, building or configuring the system, connecting it to the tools your team already uses, getting people to use it, and running it after launch. Every provider covers some of that work. Most buying mistakes come from assuming a provider covers a part it does not.

Every provider you talk to also fits one of five categories, or sells a mix of two. Each one is right for someone. Here is what each sells, where it tends to break down, and when it is the right call.

A consultant writes the plan.

Traditional consulting brings a sharp outside read on your business and a plan to act on it, usually delivered as a deck. It is strong on business context and on getting leadership to agree on a direction. You pay by the day or the week, or a fixed fee for the project.

Where it breaks down. The build is a separate project, often with someone else, and the plan ages while you look for them. The engagement ends when the deck is delivered, so what you keep is a report and a roadmap.

When it is the right call. You need a company-wide strategy, or an outside view on a merger or a reorganization.

A dev shop builds to a spec.

An AI dev shop, or an agency that works like one, brings capable engineers who build to a specification. It is strong on custom work. You pay a project fee, then change orders for anything that moves after sign-off, then a support contract if you want one.

Where it breaks down. The spec is only as good as the shop's grasp of your business, and that grasp has to come from you. Support after launch is a separate contract, when it exists. When the models or the workflow change, the system drifts, and what you keep is code someone has to maintain.

When it is the right call. You have a clear spec, a product owner and people ready to run the system once it ships.

Software handles the general case.

AI software is quick to buy and easy to trial. It works on day one for the general case, because it is built for every company and configured by you. You pay per seat or per use, billed monthly.

Where it breaks down. Setup, adoption and the fit with your workflow are your team's job. Each tool is another login and another place to check, and every seat renews whether it earns its keep or not. The vendor updates the product. Your workflow stays yours to maintain.

When it is the right call. The workflow is standard, your team has time to configure the tool, and someone owns adoption.

A hire learns the business from the inside.

Hiring puts a person, or a small team, on your payroll. You get full control, and knowledge that stays close to the business. You pay salary, benefits, recruiting and tools, plus the management time it takes to direct the work.

Where it breaks down. Hiring takes time, and one hire rarely covers all five kinds of work. A single hire still has to figure out what to build, build it alone and keep it running. When that person leaves, what they knew leaves too.

When it is the right call. AI sits at the core of what you sell, and you are building a team around it.

An operating partner installs it and runs it.

An AI operating partner finds the work worth doing, installs a system in the tools you already use and runs it for a fixed monthly price, reporting the results in dollars. What you keep is a set of systems running in your tools, reported monthly, with the partner accountable for the result.

Where it breaks down. The trade-off is fit. If a pre-built system does not match how your business works, it is the wrong choice, and a good partner will say so before you sign. Custom work outside the partner's library is priced separately. You also pay while the systems run, which is worth weighing against what each one returns.

When it is the right call. You want a few workflows working and run without building a team.

A sixth option sits beside all five: waiting. It is a real option when no workflow is costly or frequent enough to pay back yet. What waiting costs is the value of the work you keep doing by hand.

The five, side by side.

Twelve months after launch, each option leaves you holding something different, and someone different answering the phone when it breaks.

Hours sell effort and licenses sell access. A fixed monthly price sells an outcome that someone has to keep delivering. Quotes built on different models are hard to compare, so price one workflow, for one year, including the cost of running it after launch. The worksheet for that is in what AI consulting costs.

Every model has a second bill: running it.

An AI system is closer to a machine on the shop floor than a report on a shelf. It needs someone watching it, and it needs maintenance. Models are retired and replaced. Vendors change their prices. The ERP gets an update and a field moves. Every one of those events lands on someone's desk.

The quote in front of you covers the first bill: the plan, the build, the license or the salary. The second bill arrives every month after launch. It is the work of keeping a live system paying, and it comes as five jobs.

• Watch the output. Sample what the system produces, review the exceptions it sends to people, and check its work against what a careful person would have done.
• Update it as the models change. Model providers change, reprice and retire models on their own schedule. Every change gets tested against your workflow before it reaches your customers or your books.
• Keep the cost in check. Usage grows as a system proves itself. Someone routes simple work to cheaper models, caps what runs unattended and checks the bill against the value.
• Support the people using it. Answer questions, adjust approval rules as the business changes, and bring new team members up to speed.
• Report the results. Put the numbers in front of the owner every month, in dollars, against the baseline set before launch.

The jobs are small when someone owns them and expensive when nobody does. Look back at the table. Under four of the five models they fall to you by default: the consulting engagement ends with the deck, the dev shop's support is a separate contract, the software vendor maintains its product and leaves your workflow to you, and a hire carries them until the hire moves on. An operating partner's monthly price is meant to cover them, so ask any partner to show you each one in writing.

What running AI takes, month by month, is the subject of the guide to managed AI services.

Why one AI hire is rarely enough on its own.

The question owners ask most often is whether to hire someone who knows AI. Sometimes the answer is yes. For an owner-led company, though, a single hire is often a thin way to cover the work, for reasons that have little to do with the person's talent.

Start with the five kinds of work. Finding the workflows worth doing takes judgment about how your business makes money. Building the system and connecting it to your inbox, your ERP and your CRM is engineering. Getting people to use it means changing habits in the office and on the floor. Running it is operations, every month, with nobody to hand it to when that person is out. Those call for different skills, and one hire rarely covers all of them.

Then there is the work around the work. Somebody has to set the hire's priorities, judge whether what they build is any good and decide what comes next. In an owner-led company that is often the owner, who wanted the hire because AI sits outside their own field. A hire also needs what any provider needs from the business: an owner for each workflow, access to the systems, real examples and a baseline to measure against.

And people move on. A single hire learns your business over time, if they stay. When that person leaves, what they knew leaves too, along with the only full picture of how the systems were wired together.

None of this rules out hiring. It is the right call when AI sits at the core of what you sell and you are building a team around it. A software company whose product runs on AI is the clear case, and even there it is worth asking whether your engineers' weeks belong on the product or in the back office. For a distributor, a manufacturer or a field-service company that wants quotes, orders and invoices to stop eating hours, the work is a handful of workflows. Each one needs installing and then running, month after month. That takes a team, whoever employs it.

Decide in six questions.

Take them in order. The first one decides whether to spend anything at all.

1. Is there one workflow that costs enough to fix this year?

   Start from the work. List the workflows that repeat, score them on volume, the cost of a delay, the cost of an error, system access and a named owner, and see whether one stands out. The Workflow Shortlist does the scoring with you. If nothing stands out, waiting is a fair answer for now.

2. Do you need a plan for the whole company?

   A company-wide strategy, a merger or a reorganization is consulting work, and a good consultant earns the fee. If you already know which workflow you want fixed, a plan adds a step before anything changes.

3. Is the workflow standard, with someone to configure and own a tool?

   Then buy the software. Check that your team has the hours to set it up and that one person owns adoption, because setup and adoption stay with your team.

4. Do you have a clear spec, a product owner and people to run what is built?

   Then a dev shop fits. Budget the support contract and the model updates before you sign, and name the person on your side who will own the system once the shop has moved on.

5. Does AI sit at the core of what you sell?

   Then hire, and plan for a team around the first hire. Give that team a baseline for every workflow it takes on, so its results are counted the same way as anyone else's.

6. Do you want a few workflows working and run, without building a team?

   Then an operating partner fits. Ask what it installs from its library, what custom work costs, who runs each system after launch and who signs the numbers it reports.

The answers can point two ways at once. A company may buy software for one standard job and bring in a partner for the workflows that need installing and running. Write down which model owns which workflow, so nothing falls between them.

Where StrikeArc fits.

StrikeArc is an AI operating partner for the mid-market. It starts with StrikeArc Scout, the Intake & Diagnostic, which is free. StrikeArc Scout scores every play in the StrikeArc PowerPlays Library against your answers, and a person walks you through the readout, with the value at stake, the top plays and the do-not-build list. The readout is yours to keep either way.

You then choose how to own a play. The subscription is $7,500 a month, including one play installed per quarter, plus run, optimization and ROI reporting on everything live. A one-time install is $20,000 per play, installed and handed over to your team. When nothing in our library fits, Custom Discovery is $10,000, then a Custom Build and a Custom Subscription priced from the Discovery findings. Every price is on Pricing.

We install the play where your team already works: your inbox, your CRM, your accounting system. We run it, update it as the models change and keep its cost down, and the results are reported in dollars, every figure signed by a person. Every gate is a named human. Agents draft everything; they approve nothing.

If another option fits you better, we will say so on the intro call. The full side-by-side view is on Compare.

Ask every option the same questions.

Whichever way you lean, put every option through the same questions and write the answers down. The weak answers are easy to spot once they sit next to each other.

The AI Partner Scorecard holds twenty of them, from what you keep at the end to who signs the numbers, with what a strong answer sounds like and a sheet to score up to three options side by side. Bring it to every call, ours included. The questions that separate one provider from another are also in the guide to choosing an AI implementation partner.

Keep reading.