Skip to content

AI automation · Melbourne studio

AI automation use cases for small businesses, with people still in charge

Six workflows small businesses most often ask us to automate, what each one connects, where a person approves and what it deliberately doesn't do.

No payment is taken with your request. You get a written outline and quote first.

Six typical builds

What each use case does, and what it doesn't

These are typical builds that show how a first project is scoped. They are not client case studies, and there are no invented results here. Each one starts small and keeps a person in charge.

Use caseWhat it doesTools it typically connectsWhere a person approvesWhat it doesn't do
Enquiry capture, summary and routingReads new enquiries from website forms or a shared inbox, pulls out the key details (service, location, timing, budget), writes a short summary and sends it to the right person or pipeline stage.Website forms, email, a CRM or spreadsheet, team chatA person reads the summary and decides the next step. Only an agreed acknowledgement email goes out automatically, if you want one.Decide whether to take the work, price it or reply with anything beyond that acknowledgement.
Lead follow-up drafts and remindersSpots leads with no reply or decision after an agreed time, drafts a follow-up in your tone and reminds whoever owns the lead.A CRM or spreadsheet, email, calendar, team chatA person reviews, edits and sends every follow-up.Run endless automated sequences or contact anyone who hasn't contacted you first.
Quote and proposal draftingTurns a brief, call notes or form answers into a first-draft quote or proposal using your price list, templates and rules.Forms, Google Docs or Word, a price-list spreadsheet, a CRM, an AI model APIEvery figure and the final document are checked and sent by a person.Set prices on its own, or promise dates and inclusions your team hasn't confirmed.
Booking confirmations and remindersSends a confirmation when a booking is made and a reminder before the appointment, and flags reschedule requests. Much of this needs little or no AI at all.A booking tool or calendar, email or SMS provider, a CRMMessage templates are approved once. Cancellations, refunds and exceptions stay with a person.Replace your booking system or handle payment disputes.
Inbox triage and categorisationLabels incoming email by type (new enquiry, existing customer, supplier invoice, other), sets a priority and moves it to the right folder or person with a one-line summary.Gmail through Google Workspace or Outlook through Microsoft 365, a CRM, team chatAnything uncertain lands in a “needs review” folder. Replies are written or approved by a person.Delete email or reply to customers on its own.
Weekly reporting and summariesPulls the week's numbers from spreadsheets or your CRM (new leads, quotes sent, work won, overdue tasks) and writes a plain-language summary to email or team chat.Spreadsheets, a CRM, booking or accounting exports, email, team chatThe owner reads it. The summary informs decisions but makes none, and each figure links back to its source.Replace your bookkeeper or accountant, or produce financial statements.
Tools depend on your stack; each connection is confirmed in writing before we quote. Want the enquiries to land somewhere better than a spreadsheet? See custom CRM development.

Worksheet

Estimate your own time saved

We don't quote hours-saved figures we can't prove. Work out your own with this worksheet before the review; it takes about fifteen minutes with last month's inbox open.

  1. Pick one task
    Choose the one repetitive task from the table above that annoys your team most, such as sorting enquiries or chasing quotes.
  2. Count how often it happens
    Look back over the last four weeks and take the weekly average. Guessing usually overstates it.
  3. Time a few real runs
    Time three to five examples from start to finish, in minutes, including switching between tools.
  4. Mark the steps a machine could do
    Copying details, typing summaries and filing are candidates. Judgment calls are not. Estimate the share of the minutes, and be conservative.
  5. Subtract review time
    A person still checks each item. Estimate the minutes that review takes, even if it's only one or two.
  6. Do the sum
    Weekly hours saved = times per week × (minutes each × share automated − review minutes) ÷ 60.
A made-up example, to show the arithmetic only: 25 enquiries a week × (12 minutes × 60% − 2 minutes of review) ÷ 60 ≈ 2.2 hours a week. Your numbers will differ. Compare the result with the running costs below and the build quote over 12 months.
  • The task happens at least several times a week
  • It follows the same pattern most of the time
  • The information arrives in writing (forms, email, documents)
  • Mistakes are cheap to catch at the review step

Chatbots

What about a website chatbot?

It is the first thing many people picture, so it's worth being direct about it.

Our automation work focuses on what happens after an enquiry arrives: capturing it, summarising it, routing it and following it up. That is where most small businesses lose time, and where a person can easily check the output before a customer sees it.

A chatbot answers customers directly, in real time, which raises the stakes: wrong answers about prices, availability or policies go straight to the customer. For some businesses with many repetitive questions it is worth it; for many, a clear website and a fast, well-routed enquiry process do more. If you want one, describe it in your brief and we will tell you honestly whether it is a good fit.

Tools

The tools these workflows connect

We build around the tools you already use, depending on your stack. Each connection depends on the tool offering an integration or API.

  • Email: Gmail, Outlook or a shared inbox
  • Google Workspace or Microsoft 365 documents and calendars
  • Spreadsheets such as Google Sheets or Excel
  • CRMs, including off-the-shelf tools or a custom CRM
  • Booking and scheduling tools
  • Team chat such as Slack or Microsoft Teams
  • Automation platforms such as Make, Zapier or n8n
  • AI model APIs, with the account in your business's name

Control

How people stay in charge

Every build follows the same rules, agreed in writing before we start.

  1. Drafts by default
    AI steps prepare summaries, categories and drafts. Sending, pricing and committing your business stay with a person unless you decide otherwise for a specific, low-risk step.
  2. Exceptions go to a person
    When the input is unclear, incomplete or unusual, the workflow stops and asks rather than guessing.
  3. You can pause it
    Your team learns where to check each workflow and how to switch it off, so a problem never keeps running unnoticed.
  4. Only the data it needs
    The plan lists which fields are sent to an AI provider and why. Everything else stays in your systems.

Privacy

Privacy and data for Australian businesses

General information, not legal advice. Check your own obligations with a qualified adviser or the regulator.

  1. Know what leaves your systems
    The project plan lists exactly which fields are sent to an AI provider and why. Anything the workflow does not need is left out.
  2. The Privacy Act and the APPs
    Many businesses with an annual turnover of A$3 million or less are exempt from the Privacy Act, but there are exceptions, including businesses that provide a health service and hold health information (OAIC).
  3. Automated decisions must be explained from 10 December 2026
    Businesses covered by the Privacy Act will need their privacy policy to describe automated decisions that could significantly affect people's rights or interests (APP 1).
  4. Keep a person on significant decisions
    Eligibility, pricing, complaints and anything a customer would expect a human to decide stay with your team.

Pricing

How pricing works

No published price, because two workflows can differ a lot. Instead: a free review, then a written quote before anything is built.

Included in the quote

  • Business process discovery
  • Automation and integration plan
  • AI-assisted document and enquiry processing
  • Human review and exception handling
  • Testing, team handover and agreed launch support

Paid separately, directly by you

  • AI API usage, subscriptions and hosting
  • Work outside the approved scope
  • Changes to the workflow after handover, quoted before any work starts

AI providers charge per token, roughly per chunk of words, so usage cost scales with how many items you process and how long they are. AI providers usually bill in US dollars.

Software subscriptions for the automation platform, CRM or other tools are separate again. Our AI automation cost guide explains every cost line, with published Australian price ranges, and AI automation in Melbourne shows how local projects are scoped.

Honest fit

When automation isn't worth it yet

We would rather tell you now than build something you stop using.

  • The task happens a few times a month: a checklist is cheaper.
  • The process changes every week, so there is nothing stable to automate.
  • Your information lives in several places with no single source of truth.
  • Every case needs a judgement call, so the automation would only move data around.
  • A built-in feature of a tool you already pay for, such as an auto-reply or a form rule, already does the job.

What are the best AI automation use cases for a small business?

The best first candidates are repetitive tasks that follow the same pattern every time and currently involve copying, summarising or chasing: capturing and routing enquiries, drafting follow-ups, drafting quotes from a brief, booking reminders, inbox triage and weekly summaries. Pick the one that costs your team the most time each week.

Are these use cases client case studies?

No. They are typical builds that show how a first project is usually scoped, not client results. We don't publish hours-saved figures we can't back up; the worksheet on this page helps you estimate your own.

How much does AI automation cost?

SIX9 quotes each project after a free review of your workflow, because the price depends on how many tools connect and how many decisions are involved. Running costs such as AI usage and software subscriptions are paid directly by you, and we estimate them in the quote.

Will AI reply to my customers without anyone checking?

Only where you decide it should, such as a simple acknowledgement. By default we build AI steps to prepare summaries, categories and drafts, and a person approves anything that commits your business or affects a customer.

Do you build chatbots?

Our automation work focuses on what happens after an enquiry arrives: capturing it, summarising it, routing it and following it up. If you want a website chatbot, describe it in your brief and we will tell you honestly whether it is a good fit.

Which tools can you connect?

Depending on your stack: email, Google Workspace or Microsoft 365, spreadsheets, CRMs, booking tools, team chat, automation platforms such as Make, Zapier or n8n, and AI model APIs. Each connection depends on the tool offering an integration or API, so we confirm every one in writing before quoting.

Is our data used to train AI models?

That depends on the AI provider and the plan. We check the provider's current data terms with you, send only the fields a workflow needs, and recommend that the AI account is in your business's name so you control its settings.

Who owns the accounts and the workflow?

We recommend keeping the automation platform and AI provider accounts in your business's name, so you control billing and access. What is handed over at the end of the project is written into the scope.

Do we need to be in Melbourne?

No. We are based in Melbourne and work online with businesses across Australia. Reviews, testing and handover sessions happen by video call and shared documents.

Get started

Start with one workflow.
We review it for free.

Describe the task that costs your team the most time. You get a written outline, a quote and estimated running costs.

Book a free review