Inception42 and Microsoft take agentic AI out of pilots and into live government operations

G42, through its sovereign agentic AI company Inception42, and Microsoft have announced a collaboration intended to accelerate the shift from scattered, experimental AI deployments towards a single governed agent ecosystem that spans an entire organisation. By deepening interoperability between their respective platforms, the two companies want to give governments and enterprises access to a far wider pool of agents without moving data outside national borders, an arrangement that keeps sovereignty and productivity in the same operating layer rather than treating them as competing priorities.

The collaboration is being positioned as a direct contribution to the UAE's national agentic AI initiative, which has set a target of having AI agents involved in half of all federal government operations within two years. That figure signals an ambition that reaches well beyond efficiency gains inside individual departments, pointing instead towards a wholesale change in how public administration is expected to function.

The obstacle to enterprise AI is no longer capability but coherence

The rapid spread of AI agents has opened new opportunities for organisations while introducing a set of problems that surface only once deployment moves past the experimental stage. As enterprises and governments push AI into a growing number of workflows, many find themselves managing fragmented deployments, inconsistent governance, and mounting complexity across their technology environments. Requirements around security, compliance, and data sovereignty have become steadily more demanding at the same time, leaving organisations to reconcile competing pressures that rarely resolve cleanly on their own.

This collaboration addresses that tension by consolidating a single ecosystem of AI agents and everyday productivity tools within a governed, sovereign operating layer built to work at national scale. Catalyst, Inception42's enterprise and government agent operating system, is designed to run on sovereign infrastructure with all data processed in-country, and can operate on-premise, in a sovereign cloud, or in the public cloud. It works across an organisation's existing data sources to build institutional context, described as an AI Brain, meaning agents draw on the organisation's own knowledge rather than operating in isolation from it.

Interoperability, not migration, is the mechanism that makes the ecosystem work

Agents composed in Catalyst are served through Compass, Core42's sovereign model and infrastructure layer, and surface directly inside Microsoft 365 Copilot, while agents built inside Copilot compose natively back into Catalyst. Supported by Microsoft Forward Deployed Engineers, the arrangement gives governments and enterprises a shared operational framework across people, data, applications, and agents, allowing them to deploy AI with greater consistency and accountability as they move from experimentation towards production-scale adoption.

Three layers of enterprise AI that organisations increasingly need but frequently source separately now converge in a single arrangement: sovereign infrastructure, intelligent agent orchestration, and everyday user adoption. Bringing them together reduces the complexity that has slowed many AI programmes and gives organisations a trusted foundation for scaling AI across industries, functions, and teams, rather than assembling those capabilities from disconnected suppliers.

"This integration reflects how sovereignty and interoperability work together," said Ashish Koshy, CEO of Inception42, explaining that Catalyst gives governments and enterprises a comprehensive platform to build, govern, and observe their agents, and that combined with Copilot, "an agent built on Catalyst works in Copilot, and an agent built in Copilot works in Catalyst, with nothing to rebuild and no separate infrastructure to stand up." He added that demand for agents was moving faster than any single platform could meet on its own, placing Inception42 and Microsoft in a position to meet that demand together.

Sovereignty is becoming a design principle rather than a constraint bolted on afterwards

The framing matters for a sector where sovereignty and usability have often been treated as trade-offs, with organisations forced to choose between keeping data in-country and giving employees the tools they already use every day. Amr Kamel, General Manager, Microsoft UAE, said that customers were looking for AI that is "powerful, trusted, governed and practical to adopt," describing an arrangement that draws together Compass, Core42's in-country infrastructure, Catalyst's agent orchestration, and Microsoft 365 Copilot's productivity experiences into a foundation for scaling AI with confidence. By allowing agents to run with data processed in-country and integrated into everyday tools, he explained, the collaboration was helping governments and enterprises accelerate innovation while maintaining control, security, and compliance.

For the wider region, the collaboration reflects a broader shift in which governments and enterprises are seeking to build AI capability on their own terms, with the infrastructure, partnerships, and talent needed to sustain it over the long term. The significance extends past the two companies involved. As agentic AI moves from controlled pilots into the operational core of public administration, the model taking shape in the UAE offers an early template for how sovereign nations elsewhere might scale AI without ceding control of the data underneath it, a question that is likely to define enterprise and government technology strategy well beyond the Gulf.