AI can now build data tools in days. That part is getting cheaper every year.
But building is the easy part. Someone still has to check the data, control who sees it, fix it when it breaks, and show regulators and clients it is under control. AI does not remove that work. It adds to it. And under ASIC’s rules, you can outsource the work but not the responsibility.
Our view: keep one person in-house who owns your data priorities, and let a specialist team do the work.
The build is getting cheaper. The responsibility is not.
AI makes building cheap
Not long ago, a new dashboard or data feed was a project that took weeks. Now AI tools can do much of it in days. The big software platforms include AI assistants in what you already pay for.
That is good news. Paying for hours of build work makes less sense every year. As Bain puts it, AI “lowers the barrier to building software”.
AI also brings extra work
Building is the easy part. Then the real work starts:
- Checking the data is right, every day.
- Controlling who can see it.
- Fixing things when they break.
- Showing regulators and clients that it is all under control.
AI does not remove this work. It adds to it, because every new AI tool is one more thing to watch. Both of Australia’s financial regulators have seen the same gap.
APRA’s April 2026 letter looked at banks, insurers and super funds, and found them adopting AI quickly while governance fell behind. ASIC, which regulates most fund managers, saw the same thing earlier. Its 2024 review of AI use looked at 23 financial services licensees. Only about half had updated their risk policies for AI.
The responsibility stays with you
You can hire help. You cannot hand over the responsibility. ASIC’s guidance on the general obligations of a licence is plain about it:
Investors already think this way about software. FE International and PwC both report that buyers now favour software built into how a business runs, and mark down products AI could easily replace. Data work is the same. The value is not in the build. It is in running it well.
So the real question is simple. Who will do this work, and how will you stay in charge?
In-house or outsourced?
Our view: keep one person in-house who owns your data priorities, and let a specialist team do the work. Three reasons.
1. You get a whole team for less than the cost of building one. Running data well takes several skills: engineering, security, data quality, reporting, AI. One hire can’t cover them all, and several hires cost a lot.
2. Nothing stops when someone leaves. If your one data person resigns, their knowledge walks out the door. With a team behind the work, and everything written down, it stays.
3. You stay in charge. Your own person sets the direction. The team works inside your business and answers to them.
One in-house hire
- One person, one set of skills
- Knowledge sits in one head
- Holidays and resignations stop the work
- Security and audit evidence built from scratch
- A salary, whatever the workload
An in-house owner + a specialist team
- Your owner sets priorities; the team covers every skill
- Knowledge is written down and shared
- The work carries on when someone is away
- Controls and evidence already in place
- Cost moves with what you use
The numbers make the point. Your in-house owner is a cost either way. The question is what it costs to add the people who do the work.
AI can shrink the build. It does not shrink that run cost: the people checking, securing and fixing it are there every year. The full method is in our first-year cost comparison.
There is a fair exception. If you already employ someone with real spare time, and you need only one or two simple data feeds, we would not be cheaper. If you are big enough to keep a full data team busy, in-house can work. For most fund managers and VC/PE firms, it means more cost and more risk.
How Datafabric does it
This is how we work. Our engineers work as part of your team and keep your data platform running. They start from a platform already built for Australian fund managers, not from a blank page. You keep the owner in-house. We are the team.
We hold a current SOC 2 Type 1 report, and our controls are open to review in our Trust Center. Every client is priced off the same published rate card, not on hours. We use AI to get more done with every engineer, and you don’t pay for the time it saves.
You own your data and your platform. If we ever stop running it, the code and documentation pass to you, so your platform keeps going without us. Why we built it this way is in Software as a Service is broken.
Where to from here
AI will keep making the build cheaper. The firms that do best will make sure someone runs it properly, and stay in charge of it.
If you are planning a data build, or reviewing what you pay for data work today, we would be glad to compare notes.
Sources
- AI lowers the barrier to building software. Bain & Company, AI Brings Headwinds and Tailwinds to the Rule of 40, April 2026
- AI adoption across banks, insurers and super funds is outpacing governance. APRA, Letter to industry on artificial intelligence, 30 April 2026
- Outsourced functions remain the licensee’s responsibility. ASIC, RG 104 AFS licensing: Meeting the general obligations, June 2022 (updated March 2026), RG 104.34
- Buyers favour platforms with embedded workflows and proprietary data; AI-exposed products see lower multiples. FE International (Thomas Smale), SaaS Valuation Multiples in 2026, 25 August 2026
- Software buyers favour embedded workflows and high switching costs; products AI can replicate face pressure. PwC, Global M&A trends in technology, media and telecommunications: 2026 mid-year outlook, June 2026
- About half of 23 licensees had updated risk policies for AI. ASIC, REP 798 Beware the gap: Governance arrangements in the face of AI innovation, October 2024
APRA’s letter covers APRA-regulated entities (banks, insurers and superannuation trustees). ASIC’s REP 798 reviewed licensees in banking, credit, insurance and financial advice. Neither is specific to fund managers; we cite them for the pattern they describe. This article is general information, not legal advice.