Macquarie trials digital workers in HR and finance amid staff trust concerns

Digital coworkers at Macquarie: what HR needs to know now

Macquarie will pilot "digital workers" in finance and HR through an investment deal with Future Secure AI. These AI avatars will have human names and faces, work in teams, and report to a human supervisor. They use agentic AI: systems that can take on repeatable, high-volume tasks independently. The goal is faster outcomes, lower costs, fewer manual errors, and more time for strategic work.

Redesign work, don't replace people

Experts back the approach of redesigning roles, not cutting headcount. As Martin Colyer put it, the win is headspace-freeing capacity for exceptions, data stewardship and better decisions. The catch: it only works with clear guardrails, transparent success measures, and a rollback option if value isn't proven. Leaders must narrate the change, protect psychological safety, and be explicit about what stays human.

Will others follow?

Expect gradual adoption. Kathleen McAdams notes most organisations will watch early pilots, learn from mistakes, and then move. Think back to early chatbots: they started on simple queries and became standard. Digital coworkers could extend that pattern across broader workflows.

What this could change in HR and finance

First-line advice may shrink. Routine questions-holiday entitlement, policy pointers, return-to-work steps-are prime candidates for AI. That said, don't erase entry-level work entirely. Those tasks teach fundamentals and grow future HR capability.

Culture is the hard part

Introduce AI with your people, not at them. McAdams stresses collaborative change: involve employees early to reduce unease and improve adoption. Elissa Thursfield warns that ignoring concerns carries the most risk. Consult staff, and where roles shift, retain talent through retraining or redeployment-not blanket replacement.

A practical rollout checklist for HR

• Define the why: pick 2-3 high-volume, rules-based tasks with clear value (e.g., policy triage, leave calculations, data entry).
• Map the workflow: inputs, systems, exceptions, approvals, and handoffs to humans.
• Guardrails: decision boundaries, escalation rules, "what stays human," and audit trails.
• Data readiness: access controls, privacy, security reviews, and redaction of sensitive fields.
• Pilot scope: a limited population, a fixed time window, and a rollback plan.
• Success measures: accuracy, cycle time, cost per case, exception rate, employee experience (pulse surveys).
• Comms plan: why this matters, how it works, what it won't do, and where to get help.
• Training: upskill managers and HRBPs to supervise AI outputs and handle exceptions.
• Change roles deliberately: keep developmental tasks for juniors; assign higher-value work to freed capacity.
• Governance: periodical quality checks, bias testing, incident reporting, and vendor risk reviews.

KPIs worth tracking

• Accuracy of responses and data updates.
• Time-to-resolution for routine queries.
• Volume handled per FTE and per digital worker.
• Escalation/exception rates and reasons.
• Employee trust and clarity scores (short pulse surveys).
• Cost per transaction vs. baseline.

Where to start this quarter

• Select one process in HR ops (e.g., policy Q&A or absence management) and one in finance (e.g., invoice matching).
• Run a four-week shadow pilot: AI drafts, humans approve. Measure quality and speed.
• Move to supervised production for a small group. Keep weekly reviews and a clear rollback switch.

Bottom line

Digital coworkers won't fix broken processes or poor data. But with tight guardrails, a human-in-the-loop, and honest communication, they can reduce admin load and create space for deeper employee support and better decisions.

Resources

• CIPD: AI in the workplace - practical guidance
• Complete AI Training: courses by job (upskill HR teams)