The Manifesto
For decades, enterprise software has been built on one assumption: people operate software. We believe that operating model has reached its end.
I. We automated the systems
We bought software for every part of IT.
The work between the systems kept growing.
We gave IT teams applications to manage devices, licenses, identities, employees, contracts and infrastructure. We automated processes, connected systems and built workflows.
Every new application added another source of truth. Every integration added a dependency. Every automated process produced exceptions that someone still had to resolve.
Workday can be authoritative for employment while Okta is authoritative for access and Intune for device state. Each system is right about its own part. The work that matters spans all of them.
The operational burden was spread across the organization, and it never went away.
Every system automated. The work between them left to people.
II. Humans became the middleware
Someone still has to work out what is true,
decide what needs to happen, and make sure it happened.
That someone is your IT team. They reconcile records, chase exceptions, coordinate hand-offs between teams and tools, and check that the access was really closed and the laptop really came back.
We have treated this work as the cost of running IT for so long that it looks normal. It is overhead, and it grows with every person, device, application and AI agent the enterprise adds.
That is why humans became the middleware.
III. Software that performs jobs
The next generation of enterprise software
will perform the jobs it used to support.
For the first time, software can do more than follow instructions. It can understand an operational responsibility, evaluate changing conditions, decide what needs to happen within policy and carry out the work across the systems an enterprise already owns.
That changes the relationship between people, software and work. An assistant helps someone do their job. An agent is responsible for getting the job done.
The size and complexity of the enterprise should no longer decide how much manual work it takes to run it.
IV. Intelligence needs context
An agent cannot responsibly manage
what it cannot see.
Giving an AI agent access to many systems doesn't mean it understands the enterprise. Those systems hold different records, conflicting information and incomplete views of the same people, devices and software. Without continuously reconciling that evidence, an agent can confidently take the wrong action: reclaim a license from someone who still needs it, or order a device for someone who already has one.
An agent that owns an outcome needs a continuously accurate understanding of the people, devices, software, contracts, spending and policies that define the enterprise, assembled from every system that knows part of it and reconciled as it changes.
That is why XOPS maintains the Operating Position.
It is the continuously reconciled state of every person, device, application, contract and agent, assembled from source-attributed evidence across the systems that each know part of it: what is true now, where the evidence conflicts or remains unresolved, who owns it and what policy applies.
V. Responsibility requires authority
An agent that owns an outcome needs authority,
boundaries, and proof that the work was done.
Recommendations leave the work with people. Responsibility means the agent carries the work through to the end.
XOPS agents operate autonomously within the authority you define. They make decisions, coordinate actions across systems and verify that the intended outcome was achieved. Every action is governed by your policies, recorded and subject to approval where you require it. When an exception exceeds an agent's authority, it goes to the person responsible for resolving it. Governance is what lets that autonomy run in production.
VI. A new operating model
Systems of record maintain information.
Agents take responsibility for outcomes.
ServiceNow manages service records. Workday manages employee records. Identity, device, procurement and finance platforms each manage their own. Each system is designed around the information and responsibilities it manages, but enterprise outcomes routinely cross those boundaries, leaving people to reconcile the differences and coordinate the work.
The enterprise does not operate inside any single vendor's application, so responsibility for its outcomes cannot live inside one either.
XOPS gives every IT function a specialized agent that works from one shared Operating Position and acts through one governed execution layer, across the systems you already own.
Systems of record
Operating Position
Specialized agents
Governed execution
Verified outcomes
VII. Manage outcomes
Your people set the direction.
Your agents deliver the outcomes.
People define the objectives, establish the policies and set the boundaries. Agents take responsibility for delivering the outcomes, handling the work autonomously and escalating only what requires human judgment.
The point is to lift the operational burden that keeps IT professionals from the work only they can do: shaping strategy, governing risk, managing vendors and delivering for the business.
Enterprise software has spent decades helping people manage work. The next generation will take responsibility for completing it.
Meet the agents
that do the work.
