Kamiwaza launches AI platform for regulated sectors

Kamiwaza AI has launched version 1.0 of its AI orchestration platform, targeting organisations in regulated industries.

The product is aimed at sectors such as healthcare, financial services, defence and government, where moving data or widening internal access can create legal, security and compliance risks.

Kamiwaza 1.0 is built around keeping data in place rather than centralising it for AI use. The platform is intended to let organisations run AI systems across distributed data environments while maintaining existing access controls and oversight.

Three additions form the core of the release: Workrooms, hardened infrastructure built on Chainguard container images, and an upgraded version of the company's Kaizen AI agent.

Governed access

Workrooms are described as bounded collaboration spaces for teams and AI agents. Users and agents operate within their existing permissions, and each Workroom contains its own data and tools available only to authorised users.

According to Kamiwaza, those access boundaries are enforced at the platform architecture level rather than through manual policy changes or filters applied to individual agents. Every action in a Workroom is also auditable.

The structure is designed to address a common issue in regulated settings, where teams often need to work across functions on sensitive projects without opening access widely across an organisation. In practice, companies and public bodies often face a choice between restricting access so heavily that collaboration slows or loosening controls in ways that raise compliance risk.

Infrastructure layer

Kamiwaza has also built the new version on Chainguard container images, which it describes as hardened and designed to limit known software vulnerabilities that can accumulate in standard open-source container images over time.

The infrastructure uses zero-to-low CVE container images and includes software bills of materials and verifiable signatures, the company said. FIPS-ready versions are also available for federal deployments.

The focus on infrastructure reflects the scrutiny security teams place on AI systems as they move from trials into day-to-day use. In regulated sectors, concerns over patching, provenance and auditability can slow roll-outs even when organisations want to adopt AI tools.

Agent upgrade

The third major element is an updated Kaizen agent. Kamiwaza said the agent works across internal data sources through its Context Manager, so outputs draw on information held across separate systems rather than a single repository.

A new skills library lets teams define what the agent can do and under which conditions, according to the company. Kaizen has also been expanded for multi-modal analysis and output.

Kamiwaza argues that fragmented data and differing access controls have limited the usefulness of AI agents in large organisations. By connecting the agent to distributed data without requiring it to be moved into one place, the company is targeting customers that cannot adopt the common model of centralising information first and dealing with governance later.

Chief Executive Officer and Co-Founder Luke Norris said the release was shaped by those customer demands.

"Regulated industries have been clear about what they need from AI: keep our data where it is, respect our security boundaries, and give us full visibility into what the AI is doing," said Norris. "Kamiwaza 1.0 builds on the distributed data and security foundation our enterprise and government customers already rely on with hardened infrastructure, governed team collaboration, and a more capable agent that works across all of it."

Kamiwaza is entering a market in which suppliers are trying to persuade large organisations that AI can be introduced without weakening controls around sensitive records, internal systems, and classified or regulated information. For hospitals, banks, defence contractors and government departments, the question is often less whether AI models are available than whether they can be used within existing security, privacy and access rules.

Chief Technology Officer and Co-Founder Matt Wallace said the different parts of the launch were intended to address separate but related constraints.

"Workroom, Chainguard, and Kaizen each solve a distinct problem, but they add up to something bigger," said Wallace. "Teams can collaborate with AI across sensitive data without anyone-human or agent-seeing more than they should. The infrastructure they're running on has zero known vulnerabilities from day one. And the agent connecting it all understands the full context of their data rather than fragments. That combination doesn't exist anywhere else."