Privacy by design
We map what personal information each workflow touches, keep it to the minimum, and keep data in Australia where you need it to stay.
AI consulting and automation for regulated Australian businesses
Put AI into how your business runs. Safely, and with numbers to prove it.
Opsloop is a service, not software. We work alongside your team to agree the outcome and the number that proves it, redesign the process, and put the right AI tools to work, often ones you already pay for, with your people in control and the audit evidence built in.
A free 30-minute working session. You leave with a one-page brief: the process, the outcome and its number, and the risks to manage.
Led by a former CTO who has automated lending operations inside a major Australian bank and built AI that handles confidential legal work.
01Clarifymap it
02Simplifyfix it
03Amplifyautomate it
Measure. Learn. Loop again.
Most AI pilots never make it into the business.
The demo works. The rollout doesn't. Usually not because the model is weak, but because nobody redesigned the work around it.
95%
of generative AI pilots showed no measurable impact on profit and loss.
MIT Project NANDA, 2025
84%
of practitioners blamed leadership and organisational causes for failed AI projects, not the technology.
RAND Corporation
77%
of Australian mid-sized businesses already report productivity gains from AI. The rest are falling behind.
MYOB mid-market research, 2026
The gap is starting with the tool instead of the outcome, and never measuring what changed. That is the one thing Opsloop fixes.
How we're different
Build it, govern it, prove it. In one loop.
AI agencies build fast and add compliance at the end. Governance consultants write the policy and leave. Software vendors sell you a platform. Opsloop works with your team to do the change and the governance together, using whichever tools fit, and holds both to a number.
| Typical AI agency | Typical governance consultancy | Opsloop | |
|---|---|---|---|
| Starts with | A tool to build | A policy to write | The outcome, and the number that proves it |
| The process | Automated as it is | Documented as it is | Simplified before anything is automated |
| Into production | Yes | No | Yes, tested in simulation first |
| Governance | Added at the end | Documents and registers | Built into every loop: AI register entry, privacy review, audit trail |
| The result | Efficiency claims | Not measured | Baseline and agreed number, checked before go-live |
| Led by | Developers | Auditors and lawyers | A former CTO accountable for regulated operations |
- Starts withThe outcome, and the number that proves itOthers: a tool to build, or a policy to write
- The processSimplified before anything is automatedOthers: automated or documented as it is
- GovernanceBuilt into every loop: AI register, privacy review, audit trailOthers: added at the end, or documents only
- The resultBaseline and agreed number, checked before go-liveOthers: efficiency claims, or not measured
- Led byA former CTO accountable for regulated operationsOthers: developers, or auditors and lawyers
Outcome first
Start with the outcome. Attach a number. Then build.
Most AI projects start with a tool and hope for a result. We work the other way round: agree the outcome you want, put a measure on it, and record today's baseline before anything changes. Often that measure is efficiency. Just as often it is quality, speed, risk or how your customers feel.
Efficiency
Hours per file, cost per case, capacity without new hires
Speed
Turnaround time, time to decision, time to first response
Quality
Error and rework rates, first-time-right, consistency across the team
Risk and compliance
Exceptions caught, audit findings, decisions you can explain
Customer experience
Complaints, drop-off, satisfaction at the moments that matter
Your number
Whatever proves the outcome for your business. If it doesn't move, it doesn't go live.
AI governance, privacy and security
AI you can defend to your regulator, your board and your customers.
Privacy, security and governance are why most businesses hold AI back, and where most AI projects are weakest. They are where our experience sits: automating lending inside a bank, building AI for confidential legal work, and leading security at an ISO-certified software company.
45%
of mid-sized businesses name cybersecurity and data privacy as a top barrier to scaling AI.
MYOB, 2026
The rules are tightening
10 December 2026 · Privacy Act
New automated decision-making rules start. If a computer program helps make significant decisions about people using their personal information, your privacy policy has to explain how.
April 2026 · APRA
Banks, insurers and super funds were told to lift AI risk management, applying the standards they already have: CPS 230, CPS 234 and CPS 220.
Tested before it acts
Agents run in simulation on real cases first, so you see what they would do before they touch a customer or a record.
Every decision traceable
Versioned workflows and a full record of what the AI did, what data it used and who approved it. Ready for audit, and for your privacy policy.
Fits your controls
Built to sit inside the security, privacy and vendor frameworks you already answer to, so AI strengthens your certification rather than putting it at risk.
Ready for the standards your customers and auditors ask about
We build and prepare. Independent accredited bodies certify.
- ISO 27001Information security management
- ISO 42001AI management systems
- SOC 2Security and trust for enterprise and US customers
- GDPRPersonal data of UK and EU customers
- Privacy ActAustralian Privacy Principles and the new automated decision rules
- APRACPS 230 and CPS 234 for banks, insurers and super
- AI6Australia's Guidance for AI Adoption: all six practices, in every loop
Free checklist
Ready for 10 December? The Privacy Act AI checklist
[Ten] questions to find which of your systems count as automated decision-making, and what your privacy policy and controls need before the deadline.
Check your inbox.
The Privacy Act AI checklist is on its way to you@company.com.au. Not there in five minutes? Check your junk folder.
Found a process that counts?
Bring it to a free 30-minute session. You leave with a one-page brief.
Scope your first loopOne practical article a month after this. Unsubscribe any time.
How AI implementation works
One loop, done properly.
We don't sell a menu. We take one process from how it runs today to how it should run with AI, prove it with numbers, then pick the next one. Each loop builds on the last.
01 · Clarify
See how work really gets done.
A fixed-fee Diagnostic. We agree the outcome and the number that proves it, capture your key processes with the people who run them, and pick the process worth fixing first.
Fixed fee
02 · Simplify
Fix the process before you automate it.
Automating a broken process just makes it fail faster. We strip out the handoffs, rework and tools you don't need, so AI lands on something clean.
Part of the build
03 · Amplify
Ship it, with people in the loop.
We set up the AI workflow with the right tools for the job, put it into production with clear points where your people approve the work, and measure it against the number we agreed up front.
Live in production, measured
Then loop. We pick the next process, your team takes on more of the work each time, and the gains compound.
How a loop works in detailWho it's for
Growing businesses where manual work is the brake
Typically 50 to 500 people doing regulated or data-sensitive work: lending and financial services, insurance, legal, health and professional services.
- Your team spends hours rekeying data between systems that should talk to each other.
- You've run AI experiments, but none of them changed how the business works.
- Growth means hiring more people to do the same manual work.
- Privacy or compliance has stopped a good AI idea before it got started.
- Your board wants an AI plan, and you want one grounded in reality.
Results
Done from the operator's chair
Not theory. These are changes led as an executive inside the business, accountable for the result.
50%
more efficient [process, e.g. home loan processing], in [an APRA-regulated bank]
Efficiency
2x
engineering throughput, by bringing AI into how software is planned, built and shipped, at [a legal AI company]
Speed
10% → 100%
of staff using AI every day, across [sales, legal, ops and finance] in [a software company]
Adoption
APRA-regulated bank · Home lending
Two years automating home loan operations
Led the redesign and automation of home loan operations inside one of Australia's major banks.
Legal AI · Confidential client data
From SaaS seats to agent workflows
Moved sales, legal, ops and finance work off tools like Salesforce and HubSpot and onto AI agent workflows, with people directing the agents, simulation before release and every workflow versioned.
Security · ISO-certified software company
Security owned end to end
Owned security for an ISO-certified legal AI company and led a major security uplift, so law firms could trust it with their most sensitive work.
About Opsloop
Two decades of doing the work, not advising on it
Opsloop is led by a former CTO with more than 20 years in technology where getting it wrong is expensive: defence engineering in the UK, automating home loan operations inside a major Australian bank, and running technology and security for a legal AI company handling confidential client work.
We started Opsloop to do one thing well: help businesses make how they work their biggest advantage, safely.
- 20+
- years leading technology
- 2 yrs
- automating home lending inside a bank
- 1 to 80
- engineers, team built from the first hire
- 2004
- first AI research published
Insights
Practical guides from the field
Amplify
Why AI pilots stall, and how to get yours into production
The five questions to answer before you start, so the pilot has somewhere to land.Governance
AI governance for mid-sized businesses: the AI6 practices in plain English
What the government's six practices ask of you, and the minimum that works at 50 to 500 people.Privacy
The Privacy Act's automated decision rules: what changes on 10 December
How to tell whether your AI and automation count, and what to do before the deadline.FAQ
Common questions
Is Opsloop a software product?
No. Opsloop is a service. We work with your team to change how a process runs and choose the right software for it, often tools you already own like Microsoft 365, Google Workspace or your CRM. We are vendor-neutral and take no referral fees.
What is a loop?
One process taken from how it runs today to AI-assisted and measured in production. It runs in three steps: a Diagnostic to agree the outcome and its number, a redesign of the process, then the AI workflow built, tested in simulation and put live with your people in control.
What do you measure?
Whatever proves the outcome you want. Often that is efficiency, such as hours or cost per case. It can just as well be turnaround time, error rates, compliance exceptions or customer complaints. We record the baseline before anything changes.
How is Opsloop different from an AI agency or a governance consultant?
Agencies build, then add compliance. Governance consultants document, but don't build. We do both in the same loop: redesign the process, ship the AI with human checkpoints, produce the governance evidence as we go, and prove the result against an agreed number.
What happens after the first loop?
Most clients keep going. We pick the next process, your team takes on more of the work each time, and we stay on as the technology lead for the programme for as long as it's useful.
Is it safe to use AI with customer data?
It can be, if it is designed that way from the start. We keep personal information to the minimum, test agents in simulation before they act, and record what the AI did and who approved it.
Will AI put our ISO 27001 or SOC 2 at risk?
Not if it is designed in from the start. Every workflow is built to your existing controls, with access, logging, vendor risk and data handling documented, so it is ready for your next audit. We prepare; we don't certify. Certification stays with your accredited certification body or, for SOC 2, your CPA firm, which keeps the audit independent. Where you hold UK or EU personal data, we design for GDPR too.
Do the new Privacy Act rules affect us?
If you are covered by the Privacy Act and use software to make, or help make, significant decisions about people, probably. The Diagnostic shows which of your processes count and what controls they need. We work alongside your legal adviser on the privacy policy wording.
Do you work outside Melbourne?
Yes. We work remotely across Australia, with on-site workshops in Melbourne and travel where it helps.
Pick one process. Leave with a brief.
A free 30-minute working session. You get a one-page brief: the process worth fixing first, the outcome and its number, and the risks to manage. Yours to keep, whether or not we work together.