Zetaris has launched Zetaris Cloud and its AI Data Harness, aiming to solve a growing bottleneck in enterprise AI: giving agents and models secure, real-time access to fragmented business data without first centralizing it. The company is also adding AI industry veteran Steve Nouri and has joined the SpaceXAI Partner Network as a Transformation Partner, with the technologies set to be tested during the Open Agent Hackathon 2026.
Enterprise AI’s data problem is becoming harder to ignore
The latest generation of AI development tools can generate software remarkably quickly, but enterprise organizations still face a less visible constraint: getting AI systems reliable access to the data they need.
Zetaris is targeting that problem with a new AI Data Harness designed to connect AI models and agents directly to distributed enterprise data.
The company has launched Zetaris Cloud alongside the AI Data Harness, positioning the technology as a federated data layer that can work across structured, unstructured and streaming sources. Rather than requiring organizations to first copy information into a centralized repository, Zetaris says its platform allows AI systems to query data where it already resides.
That approach addresses one of the persistent challenges in enterprise AI deployment. Companies often have data distributed across public clouds, private infrastructure, databases, applications and edge environments. Bringing that information together can require extensive integration, migration and governance work.
For an AI prototype, that delay can be tolerable. For an AI agent expected to respond to current business information, it can become a major obstacle.
Zetaris argues that the industry has optimized one side of the AI development equation—building applications—while leaving the data-access problem largely intact.
Its AI Data Harness is intended to provide the missing layer between AI applications and distributed enterprise information.
A federated approach to AI data
The company’s architecture is based on accessing data in place rather than automatically duplicating it.
That distinction has implications for both governance and infrastructure costs. Data that remains in its existing environment does not necessarily need to be copied into another warehouse or lake before an AI application can use it.
Zetaris says security, policy and access controls can be applied as data is queried across different environments. The company also claims the approach can reduce total cost of ownership by up to 67%, although that figure is a company-reported estimate and will depend on individual deployments and workloads.
The strategy reflects a broader movement toward federated data architectures as enterprises attempt to make AI systems useful without undertaking massive data-centralization projects.
For AI agents in particular, data freshness matters. An agent working with outdated information can produce an apparently plausible answer while missing changes that occurred after its underlying data was copied.
Real-time or near-real-time access can therefore become an important component of enterprise agentic AI.
Zetaris Cloud is designed to make that architecture available as a cloud service while remaining independent of a particular cloud provider, data platform or AI model, according to the company.
AI agents need governed data, not just better models
The launch also comes as enterprises move from generative AI experimentation toward agent-based applications.
Large language models can generate text, code and reasoning steps, but enterprise agents need access to systems containing customer records, financial information, operational data and other business context to perform useful work.
That creates a second layer of complexity. Giving an AI system broad access to enterprise data without appropriate permissions can introduce security and compliance risks.
Zetaris is therefore emphasizing governance alongside accessibility.
The company’s pitch is not simply that AI should have access to more data. It is that AI should be able to access the right data, under existing policies, at the time it is needed.
That distinction is increasingly relevant as companies deploy AI agents capable of taking actions rather than simply answering questions.
Microsoft, Google, Amazon and Salesforce are all developing enterprise AI and agentic capabilities that connect models to business applications and organizational data. Data infrastructure providers are consequently becoming an important part of the broader AI application stack.
Hackathon puts the architecture in developers’ hands
Zetaris is planning to demonstrate its approach through the Open Agent Hackathon 2026, a 144-hour event scheduled for October 22–27 and hosted by GenAI Works.
The company says more than 5,000 developers from over 100 countries are expected to participate. Zetaris and SpaceXAI will headline the event, with developers expected to combine Zetaris’s AI Data Harness with SpaceXAI’s Grok Build to create applications using distributed enterprise data.
The hackathon is significant because it offers a practical test of the proposition: whether developers can build useful agentic applications faster when they do not first have to consolidate the data those applications need.
Zetaris has also announced that Steve Nouri, an AI industry figure and technology educator, has joined the company. His addition is intended to strengthen the company’s presence in the developer and AI ecosystem.
Data duplication becomes an infrastructure issue
Zetaris is also making an environmental argument for its architecture.
The company contends that repeatedly copying enterprise data creates additional storage, compute and networking requirements. As AI workloads expand, those duplicated datasets can increase infrastructure demand.
That issue is becoming more important as organizations build increasingly data-intensive AI applications. AI infrastructure already requires substantial amounts of computing power and electricity, making efficiency across the broader data pipeline increasingly relevant.
Zetaris says reducing unnecessary data duplication could allow more processing to happen closer to where data is created, including at the edge, instead of routing every workload through centralized data centers.
The bigger question is whether federated access can deliver the performance, security and reliability enterprises require at production scale.
If it can, the AI Data Harness concept could become an important complement to the coding and model-development tools driving the current AI boom. Enterprises may not need another way to generate software as much as they need a dependable way for that software to work with the information they already have.
Market Landscape
Enterprise AI infrastructure is increasingly developing into a multi-layer stack: foundation models and accelerators at the bottom, application and agent frameworks above them, and data-access and governance layers connecting AI systems to enterprise information.
Zetaris is targeting the latter category with a federated approach. The model competes indirectly with centralized data warehouses, data lakes, lakehouses and integration platforms while aligning with the broader movement toward real-time data access and AI-ready enterprise data.
The growth of agentic AI makes the problem more urgent. Agents require current information and permissions to act safely, meaning data governance, identity, security and real-time access are becoming core components of AI infrastructure rather than separate IT concerns.
Top Insights
- Zetaris Cloud provides a federated AI data layer designed to connect models and agents with distributed enterprise data without mandatory centralization.
- The AI Data Harness supports structured, unstructured and streaming data while maintaining security and access controls, according to Zetaris.
- Zetaris says its architecture can reduce total cost of ownership by up to 67%, although the figure remains a company claim.
- The Open Agent Hackathon will test Zetaris and SpaceXAI technologies through applications built around live enterprise data.
- Federated data access could become increasingly important as AI agents require current, governed information to execute enterprise workflows.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
