Choosing the right kind of AI consulting firm.

Most organisations shopping for AI help start by comparing firms: this proposal against that one, this day rate against that one. The bigger decision usually gets made without anyone noticing it. Before you compare firms, you are choosing a kind of firm, and the kinds are built for different jobs.

In Australia, AI consulting is sold by three broad kinds of firm: large multi-service firms, specialist data and AI consultancies, and independent contractors or small studios. Each is the right answer for some projects and the wrong answer for others. This piece sets out what each kind is built for, where each tends to fall short, and how to match the kind of firm to the work you actually have.

Three kinds of AI consulting firm.

These are tendencies, not rules. There are good and bad firms of every kind. The point is to know what each kind is set up to do well, so you can ask the right questions of the firms in front of you.

• Large multi-service firms. The big professional services firms and global systems integrators. Their strength is breadth: AI sits next to legal, risk, audit, change and technology practices, and they can staff a multi-year programme across several countries. Boards often find them reassuring. The trade-offs are that AI may be one practice among dozens, the people who sell the work are not always the people who do it, and rates reflect the brand. If the firm is also your auditor, independence rules limit the advisory work it can take on.
• Specialist data and AI consultancies. Firms whose whole business is data and AI. Their strength is depth: senior people close to the work, and experience with the data problems that decide whether an AI project lands. They tend to start faster and scope tighter. The trade-offs are a smaller bench, so check depth in your industry, and variable staying power, so check how long the firm has been delivering and whether it can support what it builds.
• Independent contractors and small studios. One person or a handful of people with a specific skill. Their strength is cost and flexibility, and they work well beside a strong internal lead who already knows what to build. The trade-offs are that one person is a single point of failure, there is no bench when they are sick or on leave, and governance, insurance and IP terms fall more heavily on you.

There is a fourth option that is not a consulting firm at all: buying an agent platform and configuring it yourself. Our piece on agent platform or AI consulting firm covers when that is the better call.

Match the kind of firm to the job.

The quickest way to choose is to describe the job honestly, then see which kind of firm it fits.

• A transformation across many countries, bundled with legal, audit or tax advice. A large multi-service firm is built for this. Few specialists can staff it, and trying to assemble it from contractors creates a coordination job of its own.
• Your first AI use cases, on data that is not yet in good shape. This is where a specialist usually earns its fee. Most of the risk is in the data, not the model, and the people who have fixed messy data before will see the problems sooner. Our piece on AI consulting vs data consulting explains why the two are hard to separate.
• An AI agent that has to work inside a regulated process. Choose for local knowledge first. APRA's CPS 230, the Privacy Act, the NDIS Practice Standards and state government AI frameworks all shape what an agent may do, and a firm that has delivered under them will design for them from the start. Our piece on the local constraints that decide whether an AI project lands goes through them.
• You already have the architecture and a technical lead, and need more hands. A contractor, or a specialist supplying embedded engineers, fits. Make sure someone on your side owns the design and the result.

Why a firm's track record in Australia matters more for AI.

AI consulting is a young label. Many firms selling it today were selling something else two years ago, and some did not exist. That is not a reason to rule a firm out, but it changes what you should check. An AI agent stands on data, integrations and governance, and those are disciplines with a long track record you can inspect.

A firm with many years of enterprise delivery in Australia brings three things that are hard to buy any other way.

• It has met the local rules before. Australian privacy law, prudential standards, sector regulators and state frameworks are not the same as their overseas equivalents. A firm that has delivered under them knows where they bite, and builds that into the plan rather than finding it at the end.
• It has references you can call. Ask for a client whose system the firm built more than two years ago and which is still running. That conversation tells you more about how a firm builds than any demo.
• It is likely to be there in year three. An agent needs owning, tuning and occasionally fixing long after go-live. A firm with a long history in the market is a safer bet to still be around when you need it, and to still have people who know what was built.

The test is simple. Ask how many years the firm has been delivering data work for Australian organisations, and ask to see something it built that is still in use. Then ask who from that original team is still with the firm.

Questions that work for any kind of firm.

Whatever kind of firm you shortlist, the same questions separate a good proposal from a weak one. Our guide on how to run the shortlist goes into more depth.

• Who will do the work, by name? Ask for the people, their experience and how much of their time is committed to you.
• What is fixed and what is open? A fixed scope with a fixed price tells you the firm has done this before. Our piece on what AI consulting costs in Australia shows how to read the quote.
• What will you own at the end? Code, prompts, data pipelines, documentation and the knowledge in people's heads. See who owns the knowledge your AI creates.
• How will we know it worked? Agree the measures before the work starts. Our guide on how to measure an AI consulting engagement sets them out.

Put the answers in your brief, so every firm responds to the same questions. Our guide on how to write an AI consulting brief shows what the brief should contain.

Why RUBIX.

RUBIX is an independent, Australian-owned data and AI consultancy, founded in Melbourne in 2011. For 15 years our work has been the data underneath AI: platforms, governance, reporting and, more recently, the agents that run on top of them. We have delivered more than 450 projects for over 115 organisations. Here is how that shows up in an engagement.

• Senior people, embedded. Our forward deployed engineers work inside your team, so the people who scope the work are the people who build it.
• Fixed scope. Our engagements are fixed-scope, so you know what you are getting and what it costs before we begin.
• Data first. We build the data foundation an agent depends on before the agent, because that is where most AI projects succeed or fail. For a leading NDIS provider we joined six disconnected systems into one governed data foundation, with a live role-based dashboard, inside eight weeks. Read the NDIS provider case study.

We are not the right kind of firm for every job. If you need a transformation across several countries bundled with audit and legal advice, a large multi-service firm is built for that. If your first AI agent has to work on data that is not yet in shape, inside Australian rules, a specialist with a long local track record is usually the better fit. Our about page sets out how RUBIX has grown since 2011.

Frequently asked questions.

General information only, not legal, financial or regulatory advice. Current as at October 2026.