how much does it cost to set up ai for a small business?
the honest answer is a range, and anyone who gives you a single number without asking what you want automated is guessing. so here are the real options, what each actually costs, and the one cost nobody puts on the invoice.
the options, with real ranges
| option | typical cost | what you get | the catch |
|---|---|---|---|
| do it yourself | free + your time | full control, you learn the tool | the hours, and the risk of a setup that never gets used |
| a course or prompt pack | $20 to $200 | templates and a method to copy | you still have to apply it to your actual business |
| a freelancer | a few hundred to a few thousand | someone builds a piece for you | quality and follow-through vary widely |
| an agency | $5,000 to $25,000+ | a full custom build and support | big spend before you know it will land |
| diagnostic-first build | a few hundred to scope, then build | a map of what to automate first, then the build | you have to be honest about your process |
the cost nobody quotes
here is the number that actually matters: most AI initiatives fail to meet expectations, and the research is blunt about why. MIT's 2025 study found 95 percent of generative AI pilots showed no measurable return, and in 2025, 42 percent of companies abandoned most of their AI initiatives, up from 17 percent the year before. the most-cited cause is not the model. it is the workflow, the redesign of how the work moves, not the tool itself.
so the cheapest option that never gets implemented is the most expensive thing you can buy. a $50 prompt pack that sits unused cost you more than a scoped build that recovers ten hours a month, because one returns nothing and the other pays for itself. price the result, not the sticker.
what actually drives the price
- scope. one workflow is a few hundred dollars of work; ten workflows is an agency project. start with one.
- done-for-you vs do-it-yourself. you are paying for someone else's hours and judgment, or spending your own.
- the mapping. the most valuable and most skipped part is knowing what to automate first. a build of the wrong process is money lit on fire, no matter how cheap the hourly rate.
the approach that keeps the first spend small
you do not need a $15,000 agency project to start, and you should not buy one until you know it will land. the lower-risk path is diagnostic-first: pay a few hundred dollars to map the single highest-volume, lowest-risk process in your business and get the build plan, then build that one thing and prove it before you expand. it is the same selection rule the businesses that succeeded used, start with high-volume, repetitive, low-risk tasks where mistakes are cheap and feedback is instant.
if you want the buyer's-guide version, what setups cost in the open market and the patterns that should make you walk, read how to hire someone to set up claude. and the full setup method is in how to set up claude for a small business.
start with the map, not the big build
the systems diagnostic is $500, the price is on the page. you get a written map of the one process worth automating first and the plan to build it, before you spend on a full build. you decide on your own schedule.
get the $500 diagnostic