What it could do, what it might do one day, how businesses like yours are supposedly doing it. But the specifics are often missing. What does an actual automation look like? What does it replace? And is it really something a small business can use, or is it still in the "big company only" basket?

The short answer is it's very much within reach. Here's a look at the real automations running right now across a range of client businesses at Strictly Savvy, the problems they replaced, and how you might adapt them for yourself.

Taming the Inbox with AI

Two of our most impactful automations right now are handling website inquiries for not-for-profit clients - and both use AI as the decision-making layer.

The first one works like this:

Every day, an automation checks the inbox, pulls through any website inquiries, and sends them to an AI tool. The AI acts as the classifier - it decides whether each email is spam or a genuine inquiry, deletes the spam, and assigns the legitimate ones to the right person along with a summary of what the inquiry is about.

The result: the person handling inquiries logs on in the morning to a clear, organised task list. No time lost to spam or information missed.

The second automation is similar but with a slightly different trigger. Instead of running at a scheduled time each day, it fires the moment an inquiry lands in the inbox. The AI classifies the type of inquiry, selects the right templated response, drafts it, and moves the email to the correct folder. The human still reviews and hits send, but the thinking and the drafting is already done.

Both automations are built this way intentionally. The automation is the engine, the AI is the brain. And right now, the combination is genuinely better at handling nuanced, variable inputs (like an email inbox) than a rules-only automation ever could be.

The Traffic Light System for AI Confidence

One question that comes up a lot when people start adding AI into automations is: what happens when the AI gets it wrong?

Our answer is a traffic light system built into the prompt.

When an automation sends information to an AI for a decision, the AI responds with one of three confidence levels. Green means it's certain - it knows exactly what the email is and what to do with it, so it goes ahead and acts. Orange means it's probably right but there's something it's not quite sure about - in that case, it flags the item for a human to review. Red means it's genuinely uncertain, and the item gets assigned to a person rather than being processed automatically.

This is especially useful if you're new to automation, handling sensitive information, or working with higher volumes where errors would be costly. You can dial the thresholds up or down depending on how comfortable you are. And as you build confidence in how the AI performs, you can loosen the rules (or tighten them) without rebuilding the whole thing.

Housekeeping Automations That Go Unnoticed

Some of the most useful automations are the ones nobody really thinks about until they disappear.

Invoice filing is a good example. Not every supplier sends invoices as a clean PDF. Some arrive in the body of an email, some as a screenshot, some in formats that don't make it obvious what they are. An automation that watches the inbox, classifies the email using AI, and then files it into the correct folder automatically means a bookkeeper never has to triage that manually again.

The same logic applies to blank draft emails. If you have ever opened an email on your phone with the best of intentions and then got distracted, you know how quickly draft folders fill up. An automation that clears blank drafts on a schedule (while leaving any actual draft with content untouched) is a small thing that saves someone a recurring job they genuinely don't want.

These automations often go unnoticed because they work. That's the point!

Sales Inquiries and Lead Capture

For one of our clients with a product-based business, three separate automations work together to keep their pipeline clean without manual data entry.

One watches the business email inbox for product inquiries and pushes them automatically into the CRM. Another captures leads from Facebook. A third processes details collected at expos and events via a simple form. All three feed into the same system, tagged by source, assigned to the right team member, and ready to action.

From there, further automations handle the immediate acknowledgement email to the customer, follow-up reminders to the team, and tasks to make sure no lead goes cold. The whole process from initial inquiry to assigned sales task runs without anyone touching it manually.

The tool doing the building here is a third-party automation platform rather than the CRM's native features, because this particular client's subscription tier didn't include all the functionality needed. That is worth knowing: what your platforms can do natively will vary depending on your plan level. It is always worth checking what your existing tools already do before adding anything external - but when the native features fall short, a well-built integration can fill the gap cleanly.

Bookkeeping Automations Worth Knowing About

Reporting is one of the biggest time drains in bookkeeping, and it is one of the most automatable.

Aged payable reports, for example. Most financial platforms can generate them, but someone still has to go in, find the report, export it, format it, and send it. An automation that pulls the data, drops it into a pre-built spreadsheet template, and sends it to whoever needs it without anyone touching it saves that time entirely.

Debtor follow-ups are another one. Most financial platforms have a built-in follow-up reminder function, but the emails they send are often generic and impersonal. An automation that sends a follow-up from the business owner's actual email address, with language that sounds like a human wrote it, at a timing you control, lands very differently. You can also layer it on top of the platform's built-in reminders, using the automation as the escalation step when the first reminders haven't landed.

What This Means for Your Business

None of these automations required rebuilding an entire tech stack. Most of them replaced a task that was already being done manually - just slowly, just expensively, and just often enough to be a drain.

If you have recurring tasks in your business that follow a consistent pattern, there is likely an automation that can handle them. The entry point doesn't have to be complicated. Start with one process, get it working, then build from there.

The best automations are often the simplest ones - the ones that run quietly in the background while you get on with the work that actually needs you.