AESP & Gibbs — AI-Native Operating Systems

Most organisations are
layering AI onto broken systems.
We redesign the system.

We design and build AI-native operating models for enterprises.
Data integrity. Workflow design. Decision architecture.

Banking & Financial Services Technology Health & Insurance Superannuation Government C-Suite Advisory
The pattern

AI is not the problem. The system is.

90% of AI agents never leave the pilot. Not because the use case is wrong. Because the system underneath it is not ready.

Across seven enterprise engagements — in banking, technology, health, superannuation, and government — the same failure points appear before a single agent is deployed. AI does not fix broken systems. It amplifies them.

01
Fragmented data

Multiple systems. Conflicting definitions. No single trusted view of the workforce or the business. AI produces outputs faster — but not reliably.

02
Unclear decision ownership

When AI influences a decision, who is accountable? Without a clear answer, decisions become inconsistent and risk accumulates silently.

03
Workflow fragmentation

Work is still structured around roles, not end-to-end workflows. AI sits on top of this. It does not resolve it.

04
The data credibility gap

There is a gap between the data organisations have and the data they trust enough to act on. AI widens this gap. It does not close it.

The AESP Operating System

A system, not a structure.

The future organisation does not adopt AI tools. It operates as a system — where data, workflow, and decisions are integrated, governed, and designed for AI from the ground up.

Layer 01

Data

One source of truth. Governed definitions. Clear ownership. If data is not trusted, decisions are not trusted. Everything starts here.

  • System of record design
  • Governed data architecture
  • Ownership and accountability
  • AI-readiness assessment
Layer 02

Workflow

Work designed around outcomes, not roles. End-to-end process architecture that eliminates fragmentation and enables automation.

  • End-to-end workflow design
  • Automation and orchestration
  • System integration mapping
  • Fragmentation removal
Layer 03

Decision

Every decision classified. Every decision owned. AI-informed, AI-assisted, or human-led — defined before deployment, not discovered after.

  • Decision classification model
  • Accountability framework
  • Governance by design
  • Audit and review mechanisms
How we work

Six ways to engage.

Scoped for enterprise conditions. Grounded in real systems, real constraints, real stakeholders.

Product 01

AI Operating System Diagnostic

A structured assessment of your data, workflow, and decision landscape. We find where the system breaks — before AI amplifies it.

4 – 6 weeks

Product 02

Applied AI Lab

Hands-on design and build of specific AI-enabled workflows. Governed, measurable, grounded in your actual systems.

8 – 12 weeks

Product 03

Operating System Design

Full operating model redesign for the AI era. Target state architecture, governance framework, and multi-horizon roadmap.

12 – 20 weeks

These are three of six defined engagement types. Each is scoped for enterprise conditions — fixed duration, tangible outputs, named accountability.

View all six products →
The risk is not failure. It is false confidence.

The risk is not that AI fails. The risk is that it succeeds.

Every quarter this goes unaddressed, the exposure compounds. AI agents are being deployed into ungoverned environments — operating on data no one has validated, influencing decisions no one has classified, running inside workflows no one has redesigned.

The risk is not that they produce errors. The risk is that they produce confident outputs from an unreliable base — and the organisation scales decisions that were never owned, reviewed, or auditable.

When the regulator calls. When the market shifts. When the three people who understood the system leave. The question will not be "why didn't we adopt AI?" It will be "why did we deploy it before the system was ready?" That question is easier to answer now than it will be in twelve months.

Our position

What we believe.

01

The Ulrich model was the right answer to the wrong era. HR Business Partners, Centres of Excellence, Shared Services — this structure was designed for a world where humans executed the volume. AI removes that assumption. The Shared Services layer is the first and largest target of automation. When it goes, the three-legged stool falls. The function needs a different architecture, not a retitled version of the existing one.

02

AI amplifies system weaknesses. It does not fix them. 90% of AI agents never leave the pilot — not because the use case is wrong, but because the data is not trusted, the processes were never documented, the workflow was not redesigned, and no one owns the decision the agent is influencing. These are not technology problems. They are operating model problems. The expertise required to fix them — people who can interpret AI outputs, challenge models, and own decisions — is the scarcest resource in most functions right now.

03

HR must be the architect of this transformation — not the first function it redesigns. AI is compressing coordination layers, removing the architecture that HR roles were built on, and shifting value from execution to judgement. The organisations that lead this are the ones that redesign before they are redesigned.

04

Organisations are buying judgement, not prompts. Anyone can deploy an agent. The value is in knowing what the agent needs to operate on — governed data, redesigned workflows, classified decisions, named owners. That combination does not exist by default. It must be designed.

Engage AESP & Gibbs

The question is not how to use AI.
The question is whether your system is ready for it.

We work with a small number of organisations at any one time. Every engagement begins with a diagnostic — not a sales process. References available to serious enquiries.

Request a conversation
The AESP Operating System — Core IP

The system that runs
below the tools.

Most AI transformations start with the tool. They should start with the operating model — the data it depends on, the workflows it sits inside, and the decisions it is meant to support.

The dominant HR operating model — business partners, centres of excellence, shared services — was designed for a world where humans executed the volume. AI is removing that assumption. The Shared Services layer is the first and largest target of automation. When it goes, the structure needs to be fundamentally redesigned, not retitled.

The AESP Operating System is a three-layer model for that redesign. It does not begin with the org chart. It begins with the data, the workflows, and the decisions — and builds the capability structure that each layer requires.

PUBLISHED FRAMEWORK

The AI-Native People Function: A Proposed Operating Model for the Post-Ulrich Era

The framework paper that extends Ulrich's 1996 model to its structural conclusion in the age of AI. Defines the five capability archetypes that replace the three-legged stool — and the conditions that must be true before AI can operate reliably inside any People function.

Dr Philip Gibbs · AESP & Gibbs Pty Ltd · © 2025 — All rights reserved

Framework visual

The model at a glance.

AESP & GIBBS · FRAMEWORK PAPER · 2025

The AI-Native People Function

A Proposed Operating Model for the Post-Ulrich Era

THE STRUCTURAL SHIFT

THE ULRICH MODEL · 1996

Three-legged stool

Business Partners
Centres of Excellence
Shared Services

AI removes the Shared Services layer — the first and largest target of automation. The stool falls.

AESP OPERATING SYSTEM · PROPOSED

Three-layer architecture

Data Layer
Workflow Layer
Decision Layer

Organised around systems and flows — not roles and reporting lines.

TWO SEQUENCES. ONE SYSTEM.

The AESP Operating System distinguishes between design sequence and implementation sequence. They are not the same thing.

DESIGN SEQUENCE

Decision → Data → Workflow

Start by defining what decisions matter, who owns them, and how AI will influence them. Then design the data architecture and workflow to serve that governance design.

BUILD SEQUENCE

Data → Process → Workflow → Decision → Operating Model

Build the foundation first. Data integrity before automation. Workflow redesign before AI deployment. Decision governance embedded before agents act.

Governance is designed first. Infrastructure is built to serve it. The sequence is not optional. It is the strategy.

LAYER ONE · TWO · THREE — THE FOUNDATION

LAYER ONE

Data

One system of record. Governed definitions. Named owners. Reconciliation as standing governance — not a quarterly exercise.

"Can we trust the data the agent will operate on?"

LAYER TWO

Workflow

Work structured around outcomes and end-to-end processes. Redesigned before AI is introduced — not automated in its current broken state.

"Has the workflow been redesigned, or just automated?"

LAYER THREE

Decision

Every AI-influenced decision classified: AI-Informed · AI-Assisted · Human-Led. Named owner. Review mechanism. Audit trail — before deployment.

"Who owns this decision when the agent gets it wrong?"

See the two-sequence model above — governance is designed first, infrastructure is built to serve it.

Capability archetypes — built above this foundation

FIVE EMERGING ARCHETYPES — A STARTING BASIS, NOT A FIXED TAXONOMY

01

People Scientists

Proprietary workforce intelligence. Own AI outputs before they inform leadership decisions.

02

Workforce System Architects

OD, change architecture, human-AI work system design — as systems engineering.

03

Performance Economists

Economics of human + AI capability combined. Reward AI-native behaviour — not just human effort.

04

AI-Native HR Technologists

Build and operate the people technology infrastructure. Engineers building a people operating system.

05

Strategic Business Partners

Embedded workforce strategists. Measured on business outcomes — not HR activity metrics.

NOTE

These archetypes represent a starting basis — not a fixed or exhaustive taxonomy. As AI capabilities develop and organisations find new ways of structuring human-AI work, additional archetypes will emerge. The framework is intentionally open to that evolution. What matters is that any archetype is grounded in the three-layer foundation.

WHERE THIS MODEL SITS

Four major firms published new HR operating models in 2025-26.

The profession is converging on similar conclusions from different directions. These are not competing visions — they are different expressions of the same structural shift. The AESP Operating System is the foundation layer that makes each of them executable.

WOWLEDGE

Most operationally detailed. Work distribution framework is genuinely useful. Built on an "access vs own" philosophy. Assumes data foundation exists.

McKINSEY

Names People Technologists as a structural pillar. Describes the target state. Does not explain the transition. Fewer than 10% of AI use cases make it past pilot.

DELOITTE

Positions agentic AI as a workforce participant. Governance principle is correct. A principle without a classification mechanism is not governance — it is intent.

MERCER

Outcome Delivery Teams with P&L accountability. Structurally bold. Transition logic missing. No sequencing guidance for organisations starting from a fragmented base.

AESP & GIBBS

The implementation layer beneath all four. Data credibility before AI. Workflow redesign before automation. Decision governance before agents act. The prerequisite layer every other model assumes but none designs.

WHAT MUST BE TRUE BEFORE THIS MODEL CAN WORK

01 DATA INTEGRITY

One source of truth. Governed definitions. Named owners. Without this, AI amplifies unreliable inputs — it does not correct them.

02 PROCESS DOCUMENTATION

Processes that live in people's heads cannot be automated — only approximated. Map and standardise before you build. Variability scales the problem, not the solution.

03 DECISION ACCOUNTABILITY

Every AI-influenced decision classified before deployment. Not reviewed after something goes wrong. Designed before agents act.

04 SEQUENCED TRANSITION

Data first. Then process documentation. Then workflow redesign. Then decision governance. Then operating model. The sequence is not optional. It is the strategy.

DESIGN BEFORE YOU BUILD

Decision classification is the first design decision — not the last implementation step. Define what decisions AI will influence, who owns them, and what the accountability mechanism is. Data architecture and workflow redesign are then built to serve that governance design. Governance is designed first. Infrastructure is built to serve it.

Dr Philip Gibbs · AESP & Gibbs Pty Ltd · © 2025 · All rights reserved

aespgibbs.com · Proposed framework — open to evolution

01 Data Foundation layer

If the data is not trusted, the decisions are not trusted.

90% of AI agents never leave the pilot. In most cases the use case is sound. The constraint is the system the agent needs to operate inside — starting with data that cannot be trusted. Most organisations do not have a single, reliable view of their workforce or business. Data lives across multiple systems with conflicting definitions, manual reconciliation, and no clear ownership.

The stakes have changed. Workforce data is no longer an internal HR reporting exercise. It has become capital markets evidence. Boards, proxy advisers, and institutional investors now scrutinise workforce composition, cost structure, pay equity, and compliance status as part of governance assessments. When a CPO presents a headcount figure, a revenue-per-employee calculation, or an underpayment exposure, they are presenting something that must be defensible — not directional.

Where data credibility fails in practice

Payroll calculates entitlements based on the employment conditions it has been told about. Rostering records hours actually worked. The employment contract governs what was agreed. When these three sources do not speak to each other — when a role change is not reflected in the payroll system, when a roster records hours against the wrong award classification — compliance failure becomes structurally likely. Under Australia's Fair Work Legislation Amendment (Closing Loopholes) Act, intentional wage theft is now a criminal offence. "We believed our systems were accurate" is not a defence that holds with regulators or boards.

The problem is rarely a shortage of data. It is a shortage of data leaders can genuinely stand behind.

What this layer addresses

System of record — one agreed source of truth for critical data, with all downstream systems drawing from the same foundation. Governed definitions — consistent meaning for key metrics across the organisation. Clear ownership — every data set has a named owner who is accountable for its quality. Lineage — the ability to trace how any output was produced, from source to decision. Reconciliation — a standing governance practice, not a crisis response.

Five questions to ask before presenting workforce data to the board

From The Data Credibility Gap, published in HRD Magazine. These questions are the foundation of the Workforce Data Integrity Framework.

01 What is our single source of truth for headcount — and does every function in this presentation agree on that source?
02 When was this data last reconciled across payroll, HRIS, and rostering?
03 Are our award and contract classifications consistent across all three systems?
04 What assumptions are embedded in this data — and have they been tested against real operational data?
05 If this data were audited tomorrow, could you reconstruct how you arrived at it?

If your revenue per employee, your headcount, or your underpayment exposure were challenged publicly tomorrow — could you defend it with confidence?

The Workforce Data Integrity Framework — Dr Philip Gibbs

The Data Credibility Diagnostic

Before any AI deployment, we assess the gap between the data an organisation holds and the data it trusts enough to act on. Eight questions. One score. A clear picture of what needs to change before AI is layered on top.

01 Do you have one agreed system of record for core data?
02 Are key metrics defined consistently across the organisation?
03 Is data reconciled across systems regularly and systematically?
04 Is there a clear owner for each critical data set?
05 Can you trace how any metric is produced, from source to report?
06 Do senior leaders trust the data enough to act without challenge?
07 Are AI tools operating on governed, validated data sources?
08 If data is wrong, would you know quickly?
32–40High credibility — AI-ready
20–31Conditional trust — foundations required
Below 20High risk — do not deploy AI at scale
02 Workflow Operating layer

Work is still structured around roles. It needs to be structured around outcomes.

Most enterprise workflows have not been redesigned in years. They were built for a world of manual handoffs and siloed systems. AI sits on top of this architecture. It accelerates individual steps. It does not fix the broken design underneath.

The workflow layer redefines how work flows — end to end, across systems, with automation embedded where appropriate and human judgement preserved where required.

Worked example — employee onboarding

Onboarding is a process every organisation runs. Most run it badly. This is what the operating system model changes.

Current state
  • Offer accepted — manual entry into HR system
  • Tickets raised separately to IT, payroll, finance
  • Multiple handoffs between teams
  • Duplicate data entry across disconnected systems
  • Delays, missed steps, inconsistent experience
  • No end-to-end visibility
AESP Operating System
  • Data layer: Single employee record created once. All systems draw from the same source.
  • Workflow layer: Onboarding defined as a single workflow, triggered at contract acceptance. Orchestrated across HR, IT, and payroll automatically.
  • Decision layer: Access provisioning is AI-assisted based on role patterns. Exceptions flagged for human review. All approvals embedded in workflow.
Outcome

Zero manual tickets raised

Outcome

Full end-to-end visibility

Outcome

Consistent, auditable experience

03 Decision Intelligence layer

If a system influences a decision, who is accountable?

AI introduces a new risk to enterprise operations. Decisions are being influenced by outputs that no one has classified, governed, or assigned ownership over. The decision layer answers this before deployment — not after something goes wrong.

The Decision Classification Model

Every decision in the operating model is classified into one of three types. Each type has a defined owner, a review mechanism, and embedded governance. No decision is deployed into an AI workflow without a classification.

Type Definition & examples Primary risk
AI-Informed
AI provides context. A human decides. Attrition risk signals. Skills gap analysis. Workforce demand forecasting. Outputs must be explainable. Decision rationale documented. Human fully accountable.
Over-reliance on patterns without business context
AI-Assisted
AI recommends options. A human approves. Candidate shortlisting. Compensation benchmarking. Workforce planning scenarios. Human signs off before execution. Clear review thresholds. Full audit trail maintained.
Automation bias. Reduced scrutiny at scale
Human-Led
AI is not appropriate. Human decides and owns. Terminations. Restructures. Employee relations cases. AI restricted to administrative support only. Governance prevents automation. Human accountability non-negotiable.
Inconsistency without structured human frameworks

The rule that governs this model

Every decision must have a defined classification, a named owner, and a clear review mechanism. If these three conditions are not met, the decision should not be automated.

04 Capability What the function requires

The Ulrich model produced three archetypes. The AI-native People function requires five — none of which exist in a conventional HR team today.

This is not a headcount reduction plan. It is a capability replacement plan. AI handles the volume that justified the previous structure. What remains must be genuinely expert, genuinely strategic, and genuinely accountable.

From The AI-Native People Function: A Proposed Operating Model for the Post-Ulrich Era — Dr Philip Gibbs, AESP & Gibbs Pty Ltd

ARCHETYPE 01

People Scientists

Build proprietary workforce intelligence that no vendor can replicate. Behavioural science + data science + productivity measurement. Own the AI outputs before they inform leadership decisions.

ARCHETYPE 02

Workforce System Architects

Design how the organisation changes. OD, change architecture, AI workflow design, and human-AI work system redesign. Own the operating model — not as a document, but as a living system.

ARCHETYPE 03

Performance Economists

Understand and optimise the economics of human and AI capability combined. Build performance frameworks that reward AI-native behaviour. Make the business case for every people investment in language CFOs trust.

ARCHETYPE 04

AI-Native HR Technologists

Build and operate the people technology infrastructure. Implement and govern the system of record and workflow layer. Build the AI intelligence layer above it. Not IT support embedded in HR — engineers building a people operating system.

ARCHETYPE 05

Strategic Business Partners

Operate as embedded workforce strategists — not HR translators. Advise Exco on workforce decisions with commercial credibility. Measured on business outcomes, not HR activity metrics. Every one embedded in the business, not in the function.

"These are not upgraded HR roles. They are a different profession. The gap between what most organisations have today and what they need is architectural — it requires building or acquiring capability that does not currently exist in the function."

The AI-Native People Function — AESP & Gibbs

Apply the framework

Ready to assess your operating model?

We begin every engagement with a structured diagnostic. Four to six weeks. A clear picture of where the system breaks and what needs to change before AI is deployed at scale.

Request a diagnostic See products →
Products

Defined scope.
Tangible outputs.
Measurable value.

Six ways to engage. Each is scoped, priced, and designed to produce a specific outcome — not a report that sits on a shelf.

Product 01
AI Operating System Diagnostic

The starting point. Most organisations do not know precisely where their operating model breaks. This diagnostic finds out — before AI amplifies the damage.

4 – 6 weeks · Fixed scope

What you get

  • Data credibility score — eight-dimension assessment with quantified gap
  • Workflow fragmentation map — where handoffs fail, where accountability disappears
  • Decision accountability audit — every AI-influenced decision classified against the AESP model
  • AI readiness rating across data, workflow, and decision dimensions
  • Prioritised remediation roadmap — sequenced, with dependencies identified
  • Diagnostic report (20–30 pages) and 60-minute executive session — decision-ready, no fluff

Why it matters

90% of AI agents never leave the pilot. Not because the use case is wrong — because the system underneath it was not examined first. This diagnostic makes the invisible visible, and gives leadership a clear decision about what to fix before committing to transformation investment.

Right for organisations where AI pilots are running but not scaling — and leadership needs to know why.

Product 02
Applied AI Lab

Hands-on design and build of specific AI-enabled workflows. Not a prototype. A governed, measurable solution operating inside your real systems.

8 – 12 weeks · Defined use case

What you get

  • Workflow redesigned first — end-to-end process architecture before AI is introduced
  • A working AI-enabled workflow deployed inside your existing systems — not a sandbox
  • Decision classification for every AI touchpoint — owned, governed, auditable
  • Governance documentation — ownership, review thresholds, audit trail
  • Baseline and post-deployment metrics tracked from day one
  • Handover pack — internal team can own and operate it without us

The output is not a report. It is something that runs.

Why it matters

The discipline here is sequencing. We redesign the workflow first. Then we embed AI into it — not the other way around. This is the difference between AI that compounds over time and AI that creates new problems at scale.

Right for organisations that have identified a high-value workflow and need it built properly — with governance, not just speed.

Product 03
Operating System Design

Full operating model redesign. We design the system the organisation needs to operate with AI embedded in every workflow and decision — from foundation to capability.

12 – 20 weeks · Enterprise scope

What you get

  • Target-state operating model — data architecture, workflow design, decision governance, capability model
  • Three-horizon transformation roadmap — sequenced priorities, dependencies identified
  • Business case for investment — built to withstand CFO and board scrutiny, linked to measurable outcomes
  • Five capability archetypes defined — who the function needs, with skills and remit specified
  • Executive and board-ready presentation materials
  • Phase 1 implementation support — we stay in the room through first-stage execution

This is not a strategy deck. It is an operating model the organisation can execute against.

Why it matters

This is the engagement for organisations that have recognised the problem is not the tools — it is the system. The output is not a strategy deck. It is an operating model the organisation can execute against, with clear owners, clear sequencing, and a commercial case that holds up to CFO scrutiny.

Right for CHROs, CFOs, and COOs who need to transform how the organisation operates — not just which tools it uses.

Product 04
AI Capability & Training

Uplifting the capability of HR and leadership teams to work with AI. Not a generic training programme. A structured engagement built around your specific use cases, your systems, and your people.

4 – 8 weeks · Modular

What you get

  • AI use case identification — structured process to surface the highest-value opportunities from inside the organisation
  • Hackathon design and facilitation — generating a real pipeline of actionable ideas, not just energy
  • Data literacy programme — HR and business leaders who can interrogate AI outputs, not just consume them
  • Workflow identification workshops — where AI fits, where it does not, and who owns the decision
  • Capability uplift roadmap — what skills the function needs and how to build or acquire them
  • Train-the-trainer model — internal capability that sustains beyond the engagement

Why it matters

Organisations have budget for AI training. Most of it is spent on generic tool awareness that produces no lasting change. This product is built around the specific operating conditions of your function — the data environment, the workflows, the decisions — so capability uplift translates directly into how work actually changes.

Right for HR and People leaders who need their teams to move from awareness to application — with measurable outcomes.

Product 05
AI-Native People Function Design

The operating model redesign for People functions facing radical headcount reduction through AI. Based on the AESP framework paper The AI-Native People Function: A Proposed Operating Model for the Post-Ulrich Era.

8 – 16 weeks · CPO-level engagement

What you get

  • Current-state capability audit — every role mapped against the AI-native model
  • Target-state function design — five capability archetypes specified with skills, remit, and headcount ranges
  • Automation impact assessment — which work AI replaces, which it augments, which it cannot touch
  • Transition roadmap — five-phase sequence from current state to AI-native operating model
  • AI governance framework — decision classification embedded across every HR workflow
  • Business case — linking the redesign to productivity, cost, and commercial outcomes
  • Board and executive-ready materials

Why it matters

If a CEO told your CPO today to run the People function on 50% of its current headcount, most functions would not know where to start. This product answers that question — not as a cost reduction exercise, but as a capability redesign. The organisations that do this deliberately will lead. Those that wait will be reorganised by someone who did not understand the function.

Right for CPOs, CHROs, and CEOs facing AI-driven workforce transformation who need to redesign the People function, not just optimise it.

Product 06
Build vs Buy Advisory

A structured assessment of whether to build AI and HR systems internally or procure enterprise platforms — and how to sequence that decision correctly in the context of your operating model and risk appetite.

3 – 5 weeks · Fixed scope

What you get

  • Capability scope assessment — what must be built, what must be bought, and what is genuinely ambiguous
  • Total cost of ownership model — 5-year TCO comparison across build, buy, and hybrid paths
  • Risk register — compliance, security, talent, and regulatory exposure across each path
  • Vendor evaluation framework — how to assess platforms against your specific operating model requirements
  • Sequencing recommendation — what to decide first, what to defer, and why the order matters
  • Executive decision paper — board-ready recommendation with clear rationale

Why it matters

The "vibe coding" moment has made building seem deceptively easy. Most organisations significantly underestimate the compliance, security, and maintenance burden of custom-built HR and AI systems. A $48M build cost versus a $35M buy-and-implement is not a technology decision. It is a strategic one — and it needs to be made with full information before a single line of code is written or a vendor is selected.

Right for CIOs, CFOs, and CPOs making significant platform investment decisions where the build vs buy question has not been properly resolved.

Not sure where to start?

Most engagements begin with a diagnostic. It establishes the baseline, identifies the priorities, and creates the foundation every subsequent product builds on.

Start with a conversation
Work

Seven engagements.
Five sectors.
One consistent pattern.

Across every engagement, the same three failure points appear before a single AI tool is deployed. The organisations differ. The system failures do not.

We work with a small number of organisations at any one time. Client relationships are confidential by default. Named case studies and references are available on request, subject to client agreement.

Sector

Banking & Financial Services

Tier 1 banks and major financial institutions. Regulated environments with complex data architectures, high compliance obligations, and executive pressure to demonstrate workforce productivity and cost discipline.

Sector

Technology

ASX-listed and global technology firms scaling through acquisition. Operating models that have not kept pace with growth. People functions running at the speed of a smaller business inside a much larger one.

Sector

Health & Insurance

Large health and insurance organisations managing complex workforce structures — including significant contingent and contracted labour pools with limited visibility across payroll, rostering, and HR systems.

Sector

Superannuation

Major superannuation funds navigating growth, regulatory change, and workforce transformation. Strategic workforce planning capability built from the ground up to support future-of-work agendas.

Sector

Government & Public Sector

State government agencies and public sector insurers. People analytics and future of work programmes requiring careful stakeholder management, governance alignment, and politically sensitive data handling.

Across all sectors

The same three failure points. Every time.

Data that cannot be trusted. Decisions that have no clear owner. AI being considered before the system underneath it is ready.

What we have delivered

Work by type.

Across seven engagements, the work has covered four consistent problem types — regardless of sector or organisation size.

Work type 01

Operating Model & AI Strategy

Designing the target-state operating model for organisations that need to function differently with AI embedded. Covers data architecture, workflow redesign, decision governance, capability requirements, and multi-horizon roadmap. Delivered at CEO, CPO, and board level. Includes business case development linked to measurable commercial outcomes.

Sectors: Technology · Banking

Work type 02

People Data & Governance

Diagnosing the gap between the data organisations hold and the data they trust. Designing governance frameworks — federated ownership models, decision rights, stewardship structures, and reconciliation practices — that make data defensible at board level. Establishing AI readiness foundations before deployment.

Sectors: Banking · Government · Technology

Work type 03

Strategic Workforce Planning

Building the capability to plan the workforce against future business scenarios — not just report on the past. Covers workforce segmentation, skills and capability mapping, scenario modelling, contingent workforce visibility, and the operating model required to sustain workforce planning as an ongoing discipline rather than an annual exercise.

Sectors: Superannuation · Health · Banking

Work type 04

People Analytics & Insights

Designing and establishing people analytics capability from the ground up. Covers analytical foundations, priority use case identification, insight infrastructure, and the operating model — including roles, governance, and data literacy — required to make analytics a sustainable function rather than a reporting team operating under a different name.

Sectors: Government · Banking · Technology

The consistent finding

What we find. Every time.

01
The data problem is always a governance problem.

Every organisation we have worked with has more data than it trusts. The issue is never a shortage of systems or reporting. It is a shortage of ownership, consistent definitions, and standing reconciliation practices. Technology does not fix this. Governance does.

02
AI is always being considered before the foundation is ready.

In every engagement, AI use cases are being scoped or piloted while the data environment that AI would operate in remains ungoverned. Deploying AI on top of this does not create value. It creates faster, more confident outputs from an unreliable base.

03
The People function is not seen as the system owner.

Data, workflow, and decisions that directly determine how the workforce operates are spread across HR, Finance, IT, and Payroll — with no single function holding end-to-end accountability. The result is coordination by exception rather than design. Everyone owns a piece. Nobody owns the system.

04
The operating model has not kept pace with the business.

Growth, acquisition, and regulatory change all outpace the operating model design underneath the People function. The structure, systems, and workflows were designed for a different version of the organisation. They have not been redesigned. They have been extended — with patches, manual workarounds, and increasingly fragile integrations.

References

References available on request.

We work with a small number of organisations at any one time and treat every engagement as confidential by default. Client references — including named contacts at CEO, CPO, and CFO level — are available to qualified enquiries.

Discretion is part of the product. Named case studies are in development and will be published with explicit client agreement.

Request a conversation
About

Dr Philip Gibbs

PhD · Chartered Psychologist · CEO & Founding Partner — AESP & Gibbs Pty Ltd

Not a generalist consultant. Not an AI vendor. A Chartered Psychologist and senior HR executive who has built a career at the intersection of people data, analytics, and enterprise operating model design — across four of the world's most demanding organisations in pharma, banking, and technology.

Two decades of applied work. Three proprietary frameworks. Author of The Data Credibility Gap, published in HRD Magazine (Australia). The founder of a firm built around what actually changes how organisations work — not what looks good in a slide deck.

AESP & Gibbs was built on a specific conviction: that most organisations are trying to solve an operating model problem with a technology decision.

The engagements with WiseTech Global and Westpac confirmed it. Different industries, different scales, different problems stated at the outset — but the same three failure points underneath every one of them. Fragmented data. Unclear decision ownership. Workflows designed around roles, not outcomes.

AI does not fix any of these. It makes them visible, faster. The work of AESP & Gibbs is to design the system that sits underneath — so that when AI is deployed, it operates in an environment built for it.

AESP & Gibbs is a principal-led firm. Dr Philip Gibbs leads every engagement. Depending on scope and sector, he draws on a curated network of senior specialist partners to bring the right expertise to the right problem. We do not carry overhead. We carry relationships.

AESP & GIBBS PTY LTD — CURRENT

Five enterprise engagements since founding. Each grounded in the same diagnostic rigour — data, workflow, decision — applied across different industries, scales, and problem types.

Westpac Group
People Data & AI Transformation — ~4 months
PROBLEMHeadcount and workforce data was being reconciled manually across multiple systems with no single owner, no governed definitions, and no clear lineage from source to board report. DELIVERABLEFederated data governance model, AI readiness assessment, target-state People Intelligence Layer architecture, three-horizon transformation roadmap with Phase 1 execution plan. OUTCOMECPO gained a defensible data foundation and a sequenced investment case. For the first time, leadership had a clear picture of what needed to change in the data environment before AI could be deployed at scale.
WiseTech Global
Operating Model & AI Strategy — ~12 months
PROBLEMA high-growth ASX-listed technology business scaling beyond 7,000 employees with a People function operating model that had not kept pace — fragmented systems, no system of record, no decision governance for AI, and an M&A integration pipeline with no scalable people architecture. DELIVERABLEAI-native People operating model, Workday system of record architecture, AI governance framework, M&A integration strategy, multi-million dollar investment business case. OUTCOMEBusiness case approved at board level. CEO and CPO had a target-state operating model the organisation could execute against — with clear sequencing, defined capability requirements, and a commercial case that held up to CFO scrutiny.
AwareSuper
Strategic Workforce Planning & Future of Work — ~5 months
PROBLEMOne of Australia's largest superannuation funds with no structured workforce planning capability. Growth and regulatory change were outpacing the function's ability to plan the workforce against future scenarios. DELIVERABLEWorkforce strategy, capability requirements framework, planning methodology, and operating model to sustain workforce planning as an ongoing discipline. OUTCOMEFund leadership gained a forward-looking workforce planning capability where none had existed — shifting from reactive reporting to scenario-based planning aligned to the fund's growth agenda.
Bupa
Contingent Workforce Diagnostic & Strategic Workforce Planning — ~9 months
PROBLEMA large health and insurance organisation with limited visibility across its non-permanent labour pool. Contingent workforce data was fragmented across payroll, rostering, and HR systems with no consolidated view of cost, risk, or workforce mix. DELIVERABLEContingent workforce diagnostic identifying structural risk and visibility gaps. Strategic workforce planning capability to inform workforce mix, cost, and capability decisions. OUTCOMEPeople Team and incoming CIO gained visibility into the contingent workforce for the first time — including risk exposure that had been invisible in existing reporting. Workforce planning capability established to support ongoing decision-making.
iCare NSW
People Analytics & Future of Work — ~9 months
PROBLEMNSW state insurer with no people analytics function. Workforce data existed but was not being used to inform strategic or operational workforce decisions. No analytical infrastructure, no defined use cases, no capability. DELIVERABLEPeople analytics programme design, analytical foundations, workforce insight priorities, and capability build to support data-led workforce decisions. OUTCOMEPeople analytics capability established from the ground up. Leadership gained structured workforce insights where previously there were none — with a defined operating model to sustain analytics as a function rather than a project.
PRIOR EXECUTIVE CAREER
National Australia Bank
Executive — People Analytics Senior executive accountability for people analytics strategy, roadmap, and capability at one of Australia's Big Four banks. Led end-to-end transformation of people data management, HR management information systems, and advanced predictive analytics — aligned to enterprise decision-making and regulatory obligations.
Commonwealth Bank of Australia
Head of Workforce Analytics Led the workforce analytics function at Australia's largest bank. Accountable for HR data governance, reporting frameworks, and analytical solutions supporting strategic and operational workforce decisions across the enterprise.
GlaxoSmithKline
Director — Organisation & People Analytics · Director — Global Product Development, QA & Insights Senior executive roles at one of the world's largest pharmaceutical companies. Built enterprise HR analytics capability across global operations. Developed the GSK wellbeing ROI calculator — referenced in the Financial Times — demonstrating 8:1 return on preventative investment. Led global resilience and organisational health programmes across 100,000+ employees.
Proprietary frameworks

Three frameworks.
One integrated system.

Each framework was developed through applied enterprise work — not theory. Together they form the intellectual foundation of every AESP & Gibbs engagement.

Framework 01

The AESP Operating System

The three-layer model for designing AI-native enterprise operations. Data integrity. Workflow design. Decision architecture. The system that organisations need to operate with AI embedded — not bolted on.

Framework 02

The Decision Classification Model

A governance model for every AI-influenced decision in the enterprise. AI-informed, AI-assisted, and human-led — each with defined ownership, review mechanisms, and accountability rules. No decision deployed without a classification.

Framework 03

The Workforce Data Integrity Framework

A three-stage model for closing the gap between workforce data held and workforce data trusted. Imagine — Investigate — Impact. Designed for CPOs and CFOs presenting data at board level, where figures must be defensible, not directional.

The team

Senior practitioners.
Not generalists.

AESP & Gibbs engagements are delivered by a small, senior team. Every person who works on your engagement is named, accountable, and operating at the level the work requires.

PG

Dr Philip Gibbs

CEO & Founding Partner

PhD, Chartered Psychologist. 20+ years at the intersection of people data, analytics, and operating model design. Former Executive — People Analytics at NAB, Head of Workforce Analytics at CBA, Director of Organisation & People Analytics at GSK. Author of The Data Credibility Gap (HRD Magazine). Originator of the AESP Operating System and the AI-Native People Function framework.

NAB · CBA · GSK · WiseTech Global · Westpac Group · AwareSuper · Bupa · iCare NSW

A DISTINCT PERSPECTIVE

What shapes this work is experience across three environments — and understanding what each one sees that the others miss. Academic research provides the rigour to challenge accepted models. Consulting engagements across multiple industries and geographies provide the breadth to recognise patterns that do not show up inside a single organisation. And sustained executive roles inside large enterprises — across banking, pharmaceuticals, and technology, in the UK, Australia, and internationally — provide the lived experience of implementation where models meet real systems and real stakeholders. Most practitioners have depth in one of these environments. This framework is built on all three.

THE EXPERT PARTNER NETWORK

AESP & Gibbs is a principal-led firm. Dr Philip Gibbs leads every engagement. There are no permanent employees — only a curated network of senior specialist partners engaged on a per-project basis.

This is a deliberate design choice. Every engagement is staffed with the right expertise for that specific problem — drawn from practitioners who have done this work inside the organisations we advise. We do not carry overhead. We carry relationships. As the firm grows, the partner network expands to bring the best available expertise to each engagement.

AS

Dr Amy Shi-Nash

Enterprise Data & AI Specialist

PhD, MBA, GAICD, Honorary Doctorate. 25 years leading enterprise-scale data, AI, and technology transformation. Former Chief Analytics and Data Officer at Tabcorp. Senior leadership roles at CBA, HSBC, and NAB. Co-founder of DataSpark. Professor and Board Member.

Tabcorp · CBA · HSBC · NAB · DataSpark

Additional specialist partners engaged by engagement type and sector.

BANKING · TECHNOLOGY · HEALTH · GOVERNMENT

Independence note. Dr Philip Gibbs is co-founder of Agile HR Analytics (AHA!), a people analytics technology platform. Prof. Amy Shi-Nash is a founder of NxtLab Innovations. All AESP & Gibbs engagements are conducted independently of both organisations and do not involve commercial recommendation of either platform.

Point of view

On AI and operating models.

The organisations getting AI right are not the ones with the best tools. They are the ones that designed the system first.

The shift happening in enterprises right now is not a technology shift. It is an operating model shift. The question is not which AI platform to adopt. The question is whether the organisation can operate with AI embedded in every workflow and decision — and whether the data, governance, and accountability structures are in place to make that safe and reliable.

Most are not. Most are experimenting with tools in environments that were never designed for them. The gap between what AI can theoretically do and what it can reliably do inside a specific enterprise is determined almost entirely by the quality of the system underneath it.

That system does not build itself. It must be designed.

The AESP position

Organisations do not need more AI tools. They need a system that allows AI, data, and human judgement to operate together with clarity and control. That system does not exist by default. It must be designed.

Thought leadership

Where the conversation is.

The debate about how organisations operate with AI is accelerating. Four major consulting and research firms published new HR operating models in 2025–26. The AESP framework engages directly with this debate — not as commentary, but as the implementation layer that makes the other models executable.

FRAMEWORK PAPER · 2025

The AI-Native People Function: A Proposed Operating Model for the Post-Ulrich Era

The paper that takes Ulrich's 1996 model to its structural conclusion in the age of AI. Defines a three-layer operating architecture — Data, Workflow, Decision — and five emerging capability archetypes as a starting basis for how the function must evolve. The implementation layer that sits beneath the Wowledge, McKinsey, Deloitte, and Mercer models — addressing the data and decision governance prerequisites each of them assumes but none designs.

Dr Philip Gibbs · AESP & Gibbs Pty Ltd · © 2025

IN PRESS · HRD MAGAZINE AUSTRALIA

The Data Credibility Gap

Why the distance between the data organisations hold and the data they trust enough to act on is the primary constraint on AI deployment in HR — more limiting than any skills gap or technology choice. With an eight-question diagnostic that organisations can run themselves.

Dr Philip Gibbs · HRD Magazine Australia · In press

LINKEDIN · APRIL 2026

The AI-Native People Function: A Proposed Operating Model for the Post-Ulrich Era

Proposes a three-layer operating system for the AI-native People function — engaging directly with Wowledge, McKinsey, Deloitte, and Mercer on the implementation gap all four models leave unaddressed. Referenced in David Green's Data Driven HR Monthly (150,000 subscribers).

Dr Philip Gibbs · April 2026 · Read on LinkedIn →

WHERE AESP SITS IN THE BROADER CONVERSATION

For how the AESP Operating System relates to the Wowledge, McKinsey, Deloitte, and Mercer models — and why the implementation layer they all assume is the work none of them designs — see the Operating System page.

View the framework comparison →
The framework — visual overview

AESP & GIBBS · FRAMEWORK PAPER · 2025

The AI-Native People Function

A Proposed Operating Model for the Post-Ulrich Era

THE STRUCTURAL SHIFT

THE ULRICH MODEL · 1996

Three-legged stool

Business Partners
Centres of Excellence
Shared Services

AI removes the Shared Services layer — the first and largest target of automation. The stool falls.

AESP OPERATING SYSTEM · PROPOSED

Three-layer architecture

Data Layer
Workflow Layer
Decision Layer

Organised around systems and flows — not roles and reporting lines.

TWO SEQUENCES. ONE SYSTEM.

The AESP Operating System distinguishes between design sequence and implementation sequence. They are not the same thing.

DESIGN SEQUENCE

Decision → Data → Workflow

Start by defining what decisions matter, who owns them, and how AI will influence them. Then design the data architecture and workflow to serve that governance design.

BUILD SEQUENCE

Data → Process → Workflow → Decision → Operating Model

Build the foundation first. Data integrity before automation. Workflow redesign before AI deployment. Decision governance embedded before agents act.

Governance is designed first. Infrastructure is built to serve it. The sequence is not optional. It is the strategy.

LAYER ONE · TWO · THREE — THE FOUNDATION

LAYER ONE

Data

One system of record. Governed definitions. Named owners. Reconciliation as standing governance — not a quarterly exercise.

"Can we trust the data the agent will operate on?"

LAYER TWO

Workflow

Work structured around outcomes and end-to-end processes. Redesigned before AI is introduced — not automated in its current broken state.

"Has the workflow been redesigned, or just automated?"

LAYER THREE

Decision

Every AI-influenced decision classified: AI-Informed · AI-Assisted · Human-Led. Named owner. Review mechanism. Audit trail — before deployment.

"Who owns this decision when the agent gets it wrong?"

Capability archetypes — built above this foundation

FIVE EMERGING ARCHETYPES — A STARTING BASIS, NOT A FIXED TAXONOMY

01

People Scientists

Proprietary workforce intelligence. Own AI outputs before they inform leadership decisions.

02

Workforce System Architects

OD, change architecture, human-AI work system design — as systems engineering.

03

Performance Economists

Economics of human + AI capability combined. Reward AI-native behaviour — not just human effort.

04

AI-Native HR Technologists

Build and operate the people technology infrastructure. Engineers building a people operating system.

05

Strategic Business Partners

Embedded workforce strategists. Measured on business outcomes — not HR activity metrics.

NOTE

These archetypes represent a starting basis — not a fixed or exhaustive taxonomy. As AI capabilities develop and organisations find new ways of structuring human-AI work, additional archetypes will emerge. The framework is intentionally open to that evolution. What matters is that any archetype is grounded in the three-layer foundation.

WHAT MUST BE TRUE BEFORE THIS MODEL CAN WORK

01 DATA INTEGRITY

One source of truth. Governed definitions. Named owners. Without this, AI amplifies unreliable inputs — it does not correct them.

02 PROCESS DOCUMENTATION

Processes that live in people's heads cannot be automated — only approximated. Map and standardise before you build. Variability scales the problem, not the solution.

03 DECISION ACCOUNTABILITY

Every AI-influenced decision classified before deployment. Not reviewed after something goes wrong. Designed before agents act.

04 SEQUENCED TRANSITION

Data first. Then process documentation. Then workflow redesign. Then decision governance. Then operating model. The sequence is not optional. It is the strategy.

DESIGN BEFORE YOU BUILD

Decision classification is the first design decision — not the last implementation step. Define what decisions AI will influence, who owns them, and what the accountability mechanism is. Data architecture and workflow redesign are then built to serve that governance design. Governance is designed first. Infrastructure is built to serve it.

Dr Philip Gibbs · AESP & Gibbs Pty Ltd · © 2025 · All rights reserved

aespgibbs.com · Proposed framework — open to evolution

Published thinking

Selected publications.

Two decades of published work across practitioner press, academic journals, and major business media — focused on people data, analytics, and organisational performance.

View full publication record on ResearchGate →

HRD MAGAZINE · AUSTRALIA

Publishing 2025

The Data Credibility Gap

How workforce data became capital markets evidence — and what that means for CPOs and CFOs presenting to boards. Introduces the Workforce Data Integrity Framework.

Sole author

Forthcoming

FINANCIAL TIMES

September 2017

Reaping the benefits of a focus on wellbeing

Referenced for development of the wellbeing ROI calculator at GlaxoSmithKline — a tool that tracked proactive versus reactive spend and demonstrated an 8:1 return on preventative investment.

FT.com

HRD CONNECT

December 2016

Getting people analytics to the organisation will help strategic decision making

As GSK's Organisation and People Analytics Director — on how HR makes its biggest organisational impact through analytics and why HR leaders must invest in talent from outside the traditional HR pool.

HRD Connect

EYEFORPHARMA

May 2016

Resilience for Change Management

On how GSK strengthens leadership and employee resilience in the face of continuous organisational change — drawing on research and applied practice across global operations.

eyeforpharma

ACADEMIC BOOKS

2010 – 2015

Selected book chapters — Edward Elgar · Palgrave Macmillan · Gower · Sage · Wiley

Contributing author across five academic volumes covering organisational wellness, employee engagement, wellbeing at work, organisational culture, and positive organisational behaviour. Published by leading international academic presses in organisational psychology and HR management.

Recognition

Selected awards & honours.

2016 · ICF

International Prism Award

International Coaching Federation — the highest honour in applied coaching. Awarded to GSK for demonstrated measurable impact from coaching programmes across the organisation.

2016 · BRITAIN'S HEALTHIEST WORKPLACE

Britain's Healthiest Employer Award

Awarded to GSK in recognition of industry-leading employer health and wellbeing strategy. Featured in the Financial Times.

2015 · GLOBAL CENTRE FOR HEALTHY WORKPLACES

Global Healthy Workplace Award — Multinational Category

Global recognition for GSK's workplace health and wellbeing programme across international operations. World's best healthy workplace designation.

2015 · BITC

Judge — Employee Wellbeing Award

Appointed as independent judge for the Business in the Community Employee Wellbeing Award — assessing organisations against leading practice in workforce health strategy.

2011 · EMPLOYEE BENEFITS

Most Effective Healthcare & Wellbeing Strategy — Organisation 1,000+ staff

Awarded to Manpower Inc. for demonstrating measurable impact from a large-scale employer wellbeing strategy across a workforce of over 1,000 employees.

Affiliations & disclosures

Other roles.

Dr Philip Gibbs is co-founder of Agile HR Analytics (AHA!), a people analytics technology platform built on the Microsoft stack for medium and large organisations. AHA! provides pre-built data models, dashboards, and AI-powered workforce insights as a software product.

AESP & Gibbs engagements are conducted entirely independently of AHA!. AESP & Gibbs does not recommend, endorse, or commercially refer clients to AHA! or any other technology vendor as part of its advisory or design work. Clients are informed of this affiliation at the outset of every engagement.

The two organisations operate in adjacent but distinct spaces. AHA! is a technology product. AESP & Gibbs designs the operating model that any technology — including analytics platforms — must sit inside to function reliably.

Contact

Start with
a conversation.

We work with a small number of organisations at any one time. Every engagement begins with a diagnostic conversation — not a sales process.

If you are a CHRO, CFO, CIO, or COO who is thinking seriously about how your organisation needs to operate with AI, we want to hear from you.

What to expect
01
Initial conversation

45 minutes. We listen. You describe the situation. We ask direct questions about data, workflow, and decision ownership.

02
Diagnostic proposal

If there is a fit, we scope a diagnostic engagement. Fixed scope, defined outputs, clear timeframe.

03
Engagement decision

You decide with full information. No pressure, no generic proposal, no follow-up deck you did not ask for.

We respond to every message. Typically within 24 hours.