Tovel helps regulated enterprises adopt AI agents with the identity, oversight and evidence their boards and regulators expect.
Boards approve pilots. Teams ship code-executing, externally connected agents. The oversight a regulator will ask for, the identity, the guardrails, the evidence, lags behind. That gap is operational risk.
Tovel closes it. We take the operational reality of production agents seriously, and we hold ourselves to the discipline we sell: claim only what we can evidence, and keep a human in the loop where it matters.
The two people building Tovel, and the delivery partner that puts it into production with customers.
Rad has spent his career in platforms, identity and security architecture across banking, insurance, telecommunications and health, delivering engineering and security tooling programmes in regulated industries. He leads AI engineering at DX1 and founded Tovel to help organisations put AI security and governance into practice. At Tovel, he leads product strategy and the architecture of its AI security and governance harness.
Alex leads DX1, the engineering consultancy formed from Innablr and DevOps1 and Tovel's delivery partner. He owns the customer relationships and the delivery engagements that put Tovel into production.
We work out of DX1's Sydney and Melbourne offices. Most demos run on video; in-person sessions are available at either office.
Sydney · headquartersMezzanine Level (North), 50 Carrington Street, Sydney NSW 2000
MelbourneLevel 1, 50 Queen Street, Melbourne VIC 3000
A production engagement with an Australian critical-infrastructure operator, framed around critical-infrastructure security obligations. This is a claim, not something you can inspect yet; the reference is published when approved.
We run our own agents inside Tovel enclaves: the same harness, identity and evidence we ship. A documented walkthrough of this implementation is available in the demo and is the proof you can inspect today.
Our positioning and maturity model track the OWASP GenAI State of Agentic AI Security and Governance report, and posture is scored against the OWASP LLM Top 10 and Agentic Security packs.
The harness is described in the terms of ASD’s agentic AI harness guidance: the layer connecting models, tools, context, state and controls, with organisational accountability. A dated control-by-control mapping to ASD's harness guidance is in preparation. Mapping is not certification or endorsement.
In a 30-minute walkthrough we demonstrate a governed workflow, explain its controls and evidence, and scope an evaluation on your AWS estate. Provisioning and live assessment follow once access and prerequisites are agreed.
After the call you receive the responsibility map, a sample receipt and a written evaluation scope.