Every week we talk to a founder who's been told they need to "implement AI." Half the time, they don't. The other half, the use-case is obvious and they've already waited too long.
This post is a plain-English breakdown of what AI automation actually is, what it realistically costs, and a simple test for whether you need it.
What AI automation actually means
"AI automation" is an umbrella term for using large language models (like Claude or GPT-4) to do work that previously required a human. That work usually falls into three buckets:
1. Responding to people — chatbots, email drafters, voice agents that answer your phone. These are the most common and often the most immediately valuable.
2. Processing information — reading documents, extracting data, summarising calls, classifying support tickets. High-volume, low-error-tolerance work that humans find tedious.
3. Triggering actions — when an email comes in, file it, update the CRM, notify the right person. "Workflow automation" that used to require Zapier + a developer.
What it costs
A rough hierarchy:
- Off-the-shelf tools (Intercom, Drift, HubSpot AI): £50–500/month. Works if your needs are standard. Limited customisation.
- Custom agents built by an agency: £3,000–15,000 to build, then hosting costs. Right for businesses with specific workflows or data.
- In-house AI team: £100k+/year. Only makes sense above a certain scale.
Most of our clients land in the middle tier — custom-built, but scope-controlled.
Three questions to test if you need it
1. Is there a task someone on your team does more than 10 times a day?
If yes, it's automatable. Whether it should be automated depends on cost vs value — but it's worth modelling.
2. Do you lose leads or customers because of slow response times?
A voice agent that picks up on the second ring and books a call is worth running the numbers on. We've seen it pay for itself in under 60 days.
3. Is there a document, form, or inbox that creates a bottleneck?
Intake forms, support tickets, invoice processing — these are the easiest wins. The AI doesn't have to be perfect; it has to be better than "no-one got to it until Tuesday."
What we tell people who fail all three tests
Stop. Do not build AI because it's exciting. Come back when the answer to one of those questions changes.
We'd rather you remember us fondly than waste your money on a chatbot nobody talks to.
If you answered yes to at least one, book a free audit. We'll map the use-cases, estimate ROI, and give you a written plan — whether you hire us or not.