There's a lot of noise around AI in business right now. Most of it is either overwhelming enterprise case studies or breathless hype about things that aren't ready for production use. This article is neither.
This is a practical list of things AI can genuinely do for a Perth SME right now — not in some theoretical future, not with a six-figure budget. Practical applications, realistic costs, starting this week if you want.
What we mean by "AI" here
For the purposes of this article, AI means large language models (LLMs) like Claude or GPT-4 — the same underlying technology behind ChatGPT. These are tools that can read and write text with human-level competency, understand context, follow complex instructions, and extract structured information from unstructured inputs.
They can be accessed through APIs and built into your existing workflows and software. You don't need a data science team. You don't need to train your own model. You need a developer who understands how to use the APIs effectively — which is exactly what boutique AI developers like Source Digital do.
10 things AI can do for your Perth business this week
1. Draft responses to enquiries and emails
Feed your AI a new customer enquiry and your typical response patterns — it drafts a personalised reply that you review and send. For businesses handling 20+ enquiries a week, this alone can save 3–5 hours. The AI doesn't send anything without your review; it just removes the blank-page problem.
2. Extract data from documents and PDFs
Invoices, contracts, application forms, technical specs — AI can read these documents and extract structured data (names, dates, amounts, part numbers) into whatever format you need. This replaces hours of manual reading and re-entry with minutes of automated processing.
3. Summarise long documents
Legal contracts, research reports, meeting transcripts, long email threads — AI can read them and produce a concise summary with the key points. For professionals billing by the hour, faster document review directly improves revenue capacity.
4. Classify and route incoming requests
New message comes in — AI reads it, determines whether it's a sales enquiry, a support request, a complaint, or something else, and routes it to the right person or queue. No human has to triage. The right person gets it immediately.
5. Generate first drafts of marketing content
Blog posts, social media captions, email campaigns, product descriptions — AI can produce solid first drafts from a brief. Not to replace your voice, but to eliminate the hours spent staring at a blank document. You edit and refine; the AI removes the starting friction.
6. Answer questions about your business using your own data
Build an AI assistant trained on your product catalog, pricing, FAQ, and policies. Your website visitors (or your own staff) can ask it questions and get accurate, contextual answers — without waiting for a human response. This is particularly powerful for businesses with large product ranges or complex pricing structures.
7. Generate reports from raw data
Give AI a spreadsheet of sales data and ask it to identify trends, anomalies, and insights. It won't replace a data analyst, but it will surface observations that would have taken hours to find manually — in seconds.
8. Automate proposal and quote generation
Combine AI with your existing pricing data and a template, and you can build a system that generates customised quotes from an enquiry brief. The AI interprets the requirements, selects the appropriate items, and populates the template. Human reviews and sends.
9. Translate technical content for non-technical audiences
Engineering specs written for engineers. Legal terms written for clients. Technical documentation written for sales staff. AI can rewrite any of these for a different audience, accurately and quickly.
10. Process and respond to reviews
New Google review comes in — AI drafts a professional, personalised response. For businesses actively managing their online reputation, this turns a 10-minute task per review into a 30-second approval.
Common thread: In every case above, the AI assists a human rather than replacing one. The value comes from removing the time-consuming, low-judgment parts of a task — leaving the human to focus on the parts that actually require judgment.
What AI can't do yet
Being honest about limitations is important:
- Complex strategic decisions — AI can inform decisions with data and analysis, but the judgment call requires a human
- Novel problem-solving — AI is excellent at pattern-matching against known situations, but genuinely new problems still need human creativity
- Highly regulated outputs — legal, medical, or financial advice that carries professional liability should always have a human in the loop
- Autonomous action at scale without oversight — agentic AI systems can take actions, but without appropriate oversight, errors compound quickly
How to start without being overwhelmed
The biggest mistake is trying to implement everything at once. Start with one application — ideally the one that addresses your most painful recurring task. Prove it works, measure the result, then expand.
A reasonable first project for most Perth SMEs:
- Pick one high-frequency, low-judgment task that currently consumes staff time
- Define what "good" looks like for the automated output
- Build a focused tool that automates that task
- Run it in parallel with the manual process for two weeks
- Validate quality, then switch over fully
Total time from decision to deployment: typically 2–4 weeks for a focused build.
Ready to find your first AI win?
Book a free 30-minute call. We'll look at your workflows and identify the highest-value AI application for your specific business.
Book Free Strategy Call