NEXESBuilt by people who love AI.
And refuse to deploy it inside a real business without a tamper-evident record of what it did, why, and who approved it.
Leads strategy, product, and the broker channel. AI developer. Sydney, Australia.
Leads operations, delivery, and customer success. AI developer. Sydney, Australia.
We love AI. We’re also paying attention.
We’re AI optimists. We see what software-defined intelligence is doing to every desk-job task and we love it — the speed, the leverage, the way one operator can now match what a whole team used to do. The pace of the last 24 months is the most exciting thing either of us has witnessed in our working lives.
But we’ve also watched friends and clients deploy AI agents into real businesses with zero recordof what those agents actually did. When something goes wrong — a wrongly-issued quote, a confidently-wrong contract clause, an inappropriate email sent at 2am — there’s nothing to point at. No audit trail. No defence. No way to tell the regulator, the insurer, or the customer what happened.
That gap is going to bite Australian businesses badly in the next two years. The Privacy Act 2026 reforms, APRA CPS 230, ASIC RG 271 — every one of them is converging on the same requirement: if you let software make a decision, you need to be able to prove how it was made.
The intelligence evidence layer.
Nexes exists to be the evidence layer underneath every AI agent running in an Australian business. Every prompt, every model output, every tool call, every email-sent / DB-write / decision-issued, every human override — all of it captured, hash-chained, retained for seven years, retrievable by you, your broker, your insurer, your auditor, or a regulator on demand.
So when AI gets it wrong — and it will — your business has the receipts to prove it wasn’t reckless. And when AI gets it right, you have the data to underwrite, defend, and scale on it. That’s the moat: not the agent itself, but the irrefutable record of what the agent did.
We don’t want AI to slow down. We want it to be accountable. Those two things are not in conflict — they are the only way the next decade of automation actually reaches the businesses that need it most.
The principles we hold ourselves to.
Dogfood before we sell
Every capability we sell, we run against our own AI workflows first. If we won’t use it, you won’t see it.
Human in the loop
Sentinel records the human review step on every high-stakes action. Approval, override, modification — all captured.
Hand off to licensed professionals
Sentinel surfaces findings. Your licensed broker, accountant, or solicitor actions them. We never pretend to replace them.
Lock you in
Month-to-month subscriptions. Full data export at any time. Your audit trail belongs to you, not us.
Give financial advice
We diagnose. We never recommend specific insurance products, financial instruments, or anything an AFSL-licensed professional should be doing.
Invent facts
Every Sentinel finding is traceable to a source document or regulatory reference. No hallucinations, no “trust the AI” black boxes.
Run the audit. See the evidence we’d generate for you.
The Sentinel Agent Cover Audit is $349 — and you only pay once you’ve seen the findings are worth it. Free intake. Pay-when-it-matters.