A one-off ChatGPT workshop won't fix your firm's AI problem, because the problem was never that your staff don't know how to use ChatGPT. It's that nothing about how your firm actually works changes once the training ends.
That's not a guess. It's the pattern showing up in nearly every serious study of corporate AI adoption over the past two years, and it's the reason "send the team to a ChatGPT training Sydney course" keeps producing enthusiastic feedback forms and flat results.
Key takeaways
- 64% of professional-services staff have received no training on using generative AI (GenAI) for their actual industry work, yet usage keeps climbing anyway — training and adoption aren't the same thing (Thomson Reuters Institute, 2025).
- Only 4% of companies report substantial value from AI; 95% of enterprise generative AI pilots show no measurable profit impact (BCG, 2024; MIT NANDA, 2025).
- Two-thirds of Australian small and medium businesses (SMBs) already use AI, but just 5% of them are "fully enabled" to get real benefit from it (Deloitte Access Economics, 2025).
- The gap between tool use and tool value is a workflow problem, not a training problem — which is why AI Smarter builds an AI Operating System (AIOS) around a firm's actual processes instead of running another prompting class.
Why doesn't a ChatGPT training Sydney course change how a firm actually works?
Generative AI (GenAI — AI models like ChatGPT that generate text, images or code from a prompt) training teaches someone to write a better prompt. It doesn't touch the five or six systems that prompt result has to pass through before it becomes billable work: your intake form, your document template, your CRM (customer relationship management system), your review process, your invoice.
Thomson Reuters Institute's 2025 Generative AI in Professional Services Report, surveying around 1,800 lawyers, accountants, tax and corporate-risk professionals, found that 64% had received no formal training on using GenAI for their specific industry work — yet the share of organisations "actively using" GenAI nearly doubled anyway, from 12% in 2024 to 22% in 2025 (Thomson Reuters Institute, April 2025). Staff are reaching for ChatGPT with or without training. What they're missing isn't prompting skill — it's a system that knows what to do with the output.
The same report found 48% of law firm professionals said their firm still has no formal GenAI policy at all. A workshop can't substitute for that either; policy and workflow are decisions about the business, not skills you can hand an individual employee in an afternoon.
What happens when firms train staff but don't change any workflow?
They get a lot of individual ChatGPT use and almost no organisational result. Boston Consulting Group's October 2024 study of over 1,000 companies found only 4% were generating "substantial value" from AI, while 74% showed no tangible value at all (BCG, 24 October 2024).
That gap widened further the following year. MIT's NANDA initiative reviewed more than 300 enterprise generative AI deployments and interviewed 52 organisations in 2025; it found roughly 95% of pilots produced no measurable impact on profit and loss (P&L — a business's income statement). Only about 5%, concentrated in integrations built around a specific workflow rather than a general-purpose chatbot, delivered significant value (reported via Fortune, August 2025, citing MIT NANDA's "GenAI Divide" research).
Read those two studies together and the pattern is hard to miss: the firms getting results aren't the ones with the most confident ChatGPT users. They're the ones who rebuilt a specific piece of work — client intake, document drafting, invoice chasing — around AI, then measured it.
Why do most Australian SMBs that use AI see no real benefit?
Australian firms are not an exception to this pattern — if anything, the gap is starker. Deloitte Access Economics, in research commissioned by Amazon and surveying more than 1,000 Australian SMBs, found two-thirds already use AI in some form, but only 5% of those users are "fully enabled to realise its potential benefits" (Deloitte Access Economics, 25 November 2025).
The same research puts a number on what's left on the table: if just one in ten Australian SMBs at the "basic" or "intermediate" stage of AI maturity moved up a single tier, it would add an estimated A$44 billion a year to the national economy. That's not a case for using AI more. Two-thirds of SMBs already clear that bar. It's a case for using it differently — with a system behind the tool, not just the tool.
For context on where that lost time actually goes: a 2024 global survey by Asana's Work Innovation Lab, covering more than 13,000 knowledge workers including in Australia, found 53% of a typical week goes to "busywork" — chasing status updates, searching for information, coordinating communication — leaving less than half the week for the skilled work people are actually paid for (Asana Work Innovation Lab). A ChatGPT licence doesn't touch that ratio. A system built around where the busywork actually happens does.
What's actually different about firms getting results from AI?
In three decades running large-scale transformation programmes at Nokia, Orange and National Australia Bank before founding AI Smarter, I never once saw a training session change a business outcome on its own. Training changed outcomes when it was paired with a redesigned process, a new system of record, and someone accountable for the result. AI is no different — it's just smaller-scale and faster to get wrong cheaply.
That's the thinking behind AIOS, AI Smarter's AI Operating System: rather than teaching staff to prompt better, it connects a firm's existing tools — CRM, email, document storage, accounting — into one system with a daily briefing and clear tiles showing what's actually moving. The point isn't a cleverer chatbot. It's that a client enquiry, a draft, or an overdue invoice now moves through a defined, AI-assisted workflow instead of depending on whichever staff member remembers to open ChatGPT that day.
What does a system look like instead of a training session?
Concretely: instead of "use ChatGPT to draft client emails," a system means the draft is triggered automatically when a new matter or file lands in the CRM, pre-filled with the client's actual details and the firm's own templates, routed to the right person for a two-minute review rather than a twenty-minute write, and logged so the firm can see how much time it saved — every time, not just when someone remembers to use the tool well.
That's the difference between a skill and an operating system. One depends on an individual's habits on a given afternoon. The other runs whether anyone remembers to open a chat window or not. It's also the difference a Sydney firm actually feels in a P&L: AI Smarter's recent work with professional-services clients shows the gains come from the second category, not the first.
How do I know whether my firm needs training or a system?
If your team already uses ChatGPT, Copilot or similar tools informally and the honest answer to "has this changed a number we track" is no, training is not the next step — a workflow audit is. Start with the three or four tasks eating the most admin time (intake, document drafting, status chasing, invoicing are the usual suspects) and ask whether AI is touching the task itself or just one person's part of it.
Sydney firms weighing up an ad-hoc course against a structured build can see how that audit works in practice on our AI automation agency Sydney page, and the full methodology behind it on the AI Smarter homepage.
A ChatGPT training Sydney session has its place — as onboarding for a system that already exists, not as a substitute for building one. If your firm is ready to find out which of its workflows are actually worth automating, book a free initial AI consultation and we'll walk through it with you.