What is a Human Systems Architect?

Why the role matters now

Agents are moving faster than operating models.

The Human Systems Architect (HSA) moves technically capable agents into a trusted, governed and valuable human-agent operating model, deciding what stays human, what can be delegated, what must be supervised and what evidence is needed before autonomy increases.

Forward Deployed Engineers (FDE) make the agent work. The Human Systems Architect (HSA) makes the agent work for the organisation.

A Human Systems Architect (HSA) is an evolution of a Service Designer - defining how humans and AI collaborate, co‑create and scale in trusted, governed and orchestrated ways.

Problem

Capability is outpacing the operating model

Customers can build impressive agents quickly, especially with strong Forward Deployed Engineer support.
Real work is full of judgement, informal practice, tacit knowledge, exceptions, handoffs and accountability.
An agent can work in a demo and still fail in the real organisation.

Technical success is not the same as operational success.

Risk

Hidden human work quietly erodes value

The hard questions become human: who notices a bad action, who can stop it, who is accountable.
Supervisors overloaded by alerts; unclear escalation; missing decision logs.
Responsible-AI principles never translated into day-to-day behaviours; benefits claimed without baselines or attribution.
Slow value leakage: manual work reappears around the agent, adoption is uneven, governance feels bolted on.

The agent works, but the organisation is not yet ready to scale autonomy.

Solution

Design the human-agent system before scaling

Map real work with jobholders, SMEs, leaders, FDEs, Consult Partners, Change, Workforce and Operations.
Define boundaries: what stays human, what is delegated, what must be supervised.
Design supervision, handoffs and escalation; activate governance in live work.
Measure value against an agreed baseline.

A hybrid process and team, deliberately designed: people, process and AI each doing the work they are best placed to do.

Not all organisations are the same

Redesign vs. Reinvention

There is a spectrum of Human Systems Architecture (HSA) work from process redesign to complete business and process reinvention.

Although both involve combining human judgement and AI-enabled automation, they differ significantly in their starting point, level of uncertainty, documentation needs, delivery approach and the degree of organizational change required. Recognizing that difference is important because it affects how work should be framed, communicated, governed and phased with clients.

Realistic to Visionary - what is your project’s goal?

Realistic

Process redesign

Process redesign applies when an organisation already has an existing, understood process and wants to improve or automate it.

Process redesign is the more familiar pattern in current HSA work definition. It focuses on taking an existing process with significant human effort and redesigning it so that AI performs more of the operational work, while humans remain in the loop for approval, exception handling, oversight and quality assurance. This is most relevant where processes are repeatable, high-volume and consistent in structure. Typical examples include banking and finance operations, HR administration, intake processes and document handling. In these settings, the methodology can be relatively standardised because the inputs, outputs, roles and control points are already broadly known and a key HSA task is to identify all the human input and value in the process.

In these cases, the underlying process was already understood at an operations level. The opportunity was not to invent a new business model, but to redesign how the work is executed, including how human input, tacit knowledge, untracked human inputs and approvals and decision points would evolve. Existing roles typically can be adapted rather than replaced by entirely new operating constructs.

Documentation can remain firmly targeted towards practical implementation.

Visionary

Process reinvention

Process reinvention applies when the opportunity is broader: the organization may need a fundamentally different way of working, often involving new roles, new capabilities and new forms of human-AI collaboration that do not map neatly onto the current process model.

Process reinvention is a broader and less predictable activity. Here the goal is not simply to improve a known process, but to rethink how a business area could operate using AI in ways that may create entirely new workflows, capabilities, or value streams. This can involve moving beyond optimization into redesigning the operating model itself. Because the future state is less familiar, the work is inherently more exploratory and requires more flexibility in both method and documentation.

My work with Kyndryl marketing work is a useful example of process reinvention. In that case, it was not initially clear where the opportunities lay, what the future process should be, or even what the right starting point for design would be. The work was therefore more conceptual and strategic, requiring exploration for understanding how the operating model might evolve.

Documentation needs to be focused on the narrative - understanding and implications. As the scale of change is often significant, business stakeholders must be able to engage with detail and clearly understand how it will operate within their existing organizational context (people, systems, governance etc.)

What the HSA does

Design the operating model

A team role: agent delivered, strategy anchored, human system designed. The HSA designs the human system that makes adoption, supervision and value real.

01Discovers how work really happens: maps workflows with the people doing the job; captures tacit knowledge, judgement criteria and baseline performance.
02Classifies work into human-led, agent-led and supervised tasks; sets automation boundaries on consequence, evidence, reversibility and absolute control gates.
03Designs decision rights and escalation: who recommends, decides, approves or vetoes; who is accountable; what decision evidence is retained.
04Designs operability: exception handling, AI-to-human handoff, validation, trust signals, supervisor visibility and operator cognitive load.
05Activates responsible AI and governance: fairness, transparency and accountability translated into cues, intervention triggers, forums, owners, logs and incident response.
06Measures value: compares live outcomes with the agreed baseline and separates the agent contribution so benefits are attributable and auditable.

What customers gain

Value with safer scale

A deployable operating model: a system people know how to use, supervise, govern and improve before it launches.
Higher adoption and trust: people know what the agent does and how to intervene.
Safer delegation: clear boundaries, escalation routes, named accountability and evidence thresholds.
Lower hidden cost: less unmanaged monitoring, fewer handoffs, reduced supervisor overload.
Better decisions, stronger case: human judgement and industry expertise alongside AI, with agreed baselines and auditable results.
Risks reduced: failed adoption, unsafe autonomy, unclear accountability and governance drift.

Links