Should AI readiness make your firm pause? The evidence says no.

When a firm realises its documents are scattered, its know-how lives in a few people's heads and nobody owns the AI question, the natural reaction is to pause: "Let's get our house in order first, then look at AI."

It sounds prudent. But a pause on approved projects does not establish that informal use has stopped.

A pause may leave informal AI use unaddressed.

Thomson Reuters' 2026 Future of Professionals report, a survey of 1,816 legal, tax and compliance professionals across 62 countries, found that 74% already use AI tools every week. One in three lawyers, accountants and compliance professionals said they use AI their organisation hasn't approved.

The detail that matters most for a firm considering a pause: that figure rises to 41% among people who say their organisation is moving too slowly on AI.

The survey shows an association between perceived organisational delay and unapproved use, not proof that delay causes it. It is a global survey, not an Australia-specific result. The Australian evidence separately shows widespread use of general-purpose tools: in the Victorian Lawyer Census, 65% of lawyers using AI were using general-purpose tools like ChatGPT, Claude, Gemini and Copilot.

A pause alone does not resolve the risk or give you visibility of current use.

The regulators aren't asking you to wait

The Queensland Law Society's guidance is explicit that it does not seek to discourage AI use. What it asks for is guardrails: approved tools only, clear conditions of use, partners accountable for the output, and real scrutiny of how a provider handles your data.

Those are practical conditions to work on, with timing depending on the firm and the proposed use. There is also evidence of an association with perceived benefits. The Victorian census found that exposure to training and workplace guidelines was associated with lawyers recognising more benefits from AI, without any increase in how risky they perceived it.

So what does readiness actually change?

Readiness doesn't decide whether you use AI. It decides which AI you can trust, and when.

It helps to separate two kinds of use:

• Self-contained use: drafting from scratch, summarising a document someone hands the tool, research, tidying correspondence. This depends mostly on having a sanctioned, suitable tool and clear rules about what can go into it, along with an informed reviewer. The time needed depends on the task, information and approvals.
• AI that works across your firm's own information: searching matter files, drafting from your precedent bank, answering questions from your know-how. This depends on the foundations: one reliable home for documents, consistent naming, knowledge that has been written down. This genuinely waits on readiness work.

The Australian data shows why the second kind takes longer. LEAP's 2026 research found 68% of Australian respondents work across three or more platforms, and a third have no documented process for capturing knowledge when staff leave. No tool fixes that by itself.

The answer is to sequence: assess a bounded use against its actual conditions, and build the wider foundations alongside it. Some uses should wait until their prerequisites are resolved.

Who firms are turning to

Firms looking for help in 2026 tend to land in one of four places:

• Legal technology vendors. Practice management and legal AI providers are building AI into their platforms, and several offer free readiness checks. These can be useful, but they naturally point towards the vendor's own product.
• IT providers and AI consultancies. Strong on systems and implementation, though not always grounded in a law firm's professional obligations.
• Large law firms' AI practices. Deep expertise in AI governance, mostly directed at corporate and government clients.
• Law societies. Free guidance, policy templates and CPD. Essential reading, but deliberately general.

What's often missing for small and mid-sized firms is an independent view of the firm itself: where its foundations are strong, where they're weak, and what that means for which AI it can use now and which needs work first.

What to do instead of pausing

1. Bring current AI use into the open. Ask your people what they're using and for what, without penalty for honest answers.
2. Set short, clear ground rules. Which tools are approved, what can and can't go into them, who is accountable. Your law society's guidance and templates are a good starting point.
3. Choose a business-grade tool for self-contained work. One whose terms keep your client data out of model training.
4. Start the foundation work alongside it. Pick one practice area, give its documents a single home, and start capturing the know-how of the people you'd least like to lose.
5. Test as you go. Try AI on that one cleaned-up area. You'll learn what the foundations need faster than by planning in the abstract.

Readiness isn't a gate you wait behind. It's the map that tells you what you can do now, what can run alongside it, and what is worth building towards.