Who can actually run AI for your business?
There are six common ways to put AI to work in a small business. Here's what the research says about how these projects go, and what each option actually gives you.
What The Research Says
Most AI projects stall. Who builds it matters.
How often AI projects succeed
MIT studied hundreds of company AI projects. In-house builds succeeded about a third as often as projects bought from specialists.
How often AI-written code failed security checks
Veracode tested more than 100 AI models on 80 coding tasks. Bigger models did no better.
What an in-house hire costs
Median U.S. software developer pay is $135,980. Benefits make up about 30% of what employers spend on pay, which adds roughly $58,277.
Of organizations whose AI was breached, 97% had no proper access controls on it.
Gartner expects over 40% of AI agent projects to be canceled by 2027, citing cost, unclear value, and weak risk controls.
Six Ways To Do It
What each option actually gives you.
Swipe to see all six.
BrewTech + Swirls
BrewTech runs it; Swirls powers it
- YesWorks with your toolsBrewTech already manages your systems; Swirls connects to hundreds of apps
- YesTies your tools togetherOne system across your apps, built in Swirls
- YesAsks before it actsImportant actions wait for your approval
- YesYou own itWritten down plainly in Swirls and yours to keep
- YesMaintained for youBrewTech monitors and maintains it
Built-in IT software AI
AI inside tools like Syncro, HaloPSA, SuperOps
- PartlyWorks with your toolsOnly sees data inside that one product
- NoTies your tools togetherCan't reach your other apps
- YesAsks before it actsGood approval controls
- NoYou own itLives inside the vendor's product
- PartlyMaintained for youThe vendor maintains the software, not your setup
DIY AI assistants
ChatGPT, Claude and similar, set up yourself
- PartlyWorks with your toolsConnects to many apps, one person at a time
- PartlyTies your tools togetherLinks apps inside a single chat
- PartlyAsks before it actsAsks before acting, inside the chat
- NoYou own itChats and prompts, not a system
- NoMaintained for youYou maintain it yourself
AI consultants
One-off custom builds
- PartlyWorks with your toolsHas to be given access first
- YesTies your tools togetherCustom builds can link apps
- PartlyAsks before it actsDepends on who you hire
- YesYou own itYou usually get the code
- NoMaintained for youOften gone after the project
Automation apps
Zapier, n8n and similar
- YesWorks with your toolsThousands of app connections
- YesTies your tools togetherLinking apps is what they do best
- NoAsks before it actsFew built-in approval steps
- PartlyYou own itYours, but setups get hard to follow
- NoMaintained for youYou set up and fix it yourself
In-house AI hire
An employee building systems with AI chatbots
- PartlyWorks with your toolsHas access, but connecting systems is learned on the job
- PartlyTies your tools togetherPossible, but often fragile, chatbot-written code
- NoAsks before it actsApprovals and audit trails are rarely built in
- PartlyYou own itYours, but it lives in one person's head
- PartlyMaintained for youStops when that person is out or leaves
Yes, Partly, and No are our assessment of each option as it's usually delivered today.
What About Hiring Someone?
Same story as in-house IT. Higher stakes.
Plenty of businesses hand AI to one employee and a chatbot. The chatbot will happily write the code. It won't tell you what it missed about your data, your security, or what breaks six months from now.
- About $194,000 a year in salary and benefits for one person learning on the job
- Data, security, and compliance figured out as they go
- Chatbot-written code that nobody else reviews
- No backup when they're out sick, on vacation, or gone
- When they leave, the know-how leaves with them
- A team that builds and runs these systems every day
- Security and data handling are part of the build from day one
- Every system is written down plainly in Swirls and reviewed
- Monitored and maintained, by people who already know your IT
- You own what we build, whoever is on your payroll
Your in-house people still matter. They know the business best. We work with them, so they spend their time on the work, not on keeping the systems running.
See what it would look like for your business.
Bring one process that eats your team's time. We'll tell you honestly which of these options fits it best, even if it isn't us.
Charts and figures come from published research. The Yes, Partly, and No ratings are BrewTech's assessment, based on public product information and our own client projects. BrewTech builds and runs every system on Swirls. Think we got something wrong? Tell us.