Right now, software vendors slap the word "AI" on everything, including basic workflows that have existed for fifteen years. That makes it hard for a business owner to know what they are actually buying.

Here is the simplest rule I use during AI consulting calls: if the job follows clear, repeatable steps, use plain automation. Only bring in AI when the job requires reading, interpreting, or writing human language.

Brass mechanical clockwork gears and a telegraph relay arm wired to a glowing amber glass vacuum tube in a brass cage on a workshop bench
Predictable clockwork automation handles repeatable steps on the left; an AI model steps in on the right only where messy human language needs reading.

When plain workflow automation wins

Deterministic automation takes A and turns it into B using fixed rules. It does not get creative, and it does not make things up. In my automation work, plain workflows usually start around $900 and handle jobs like:

  • Bookings and payments: On Don-O-Ray Farm Adventure, online reservations, payments, and gate passes flow end to end without staff copying orders by hand.
  • Event registrations and member syncs: Les Compagnons des francs loisirs runs event calendars, online sales, and volunteer communications from one bilingual system instead of scattered spreadsheets.
  • Scheduled alerts and monitoring: Checking whether linked school websites are down and emailing staff before visitors hit broken links.

When AI actually earns its keep

Plain automation breaks down as soon as the input is messy human language. That is where an AI model earns its keep (typically starting around $1,500 for an assistant wired into your site or workflow):

  • Answering open-ended visitor questions: Visitors rarely phrase questions the way your navigation menu does. On ÉcolesOntario.ca, the custom chatbot reads live directory records to answer parents in plain French.
  • Reading unstructured emails or PDFs: Extracting customer details, dates, and requested items from free-form emails or varied supplier invoices.
  • Completing and verifying messy datasets: Using an automated AI crawl to research and fill missing fields across hundreds of spreadsheet rows before importing them into a site.

The best systems combine both

You do not have to guess which one your bottleneck needs before calling. During a free Digital Efficiency Assessment, we walk through the exact task, and I will tell you honestly if plain automation solves it for half the cost.