Personalisation Platform Lead · Agentic marketing
I build agentic systems that connect intelligence, personalisation and orchestration to measurable business outcomes.
Personalisation at enterprise scale is becoming an agentic orchestration problem. I design the systems that enable AI agents to understand customers, make decisions, coordinate actions across channels, and continuously optimise outcomes under enterprise-grade governance and controls.
- 16+ yrs
- Personalisation, MarTech & AI enablement
- AEP
- RT-CDP · AJO · CJA · AEM · Target
- Multi-agent
- Systems designed, built and evaluated hands-on
The agentic shift
Marketing is moving from tools to agents
Wave 1
Point solutions
Siloed teams and siloed tools. Every channel optimises for itself.
Wave 2
Platform consolidation
One data foundation, one shared profile. Most enterprises are here, and their operating model has not caught up.
Wave 3 · Now
Agentic orchestration
Agents build audiences, assemble journeys and generate creative from a single brief. People set intent and approve.
Once execution is automated, the differentiator is end-to-end judgement across identity, audience, journey and measurement. That is the layer I work in.
Capabilities
End-to-end ownership of the agentic marketing capability
Fifteen responsibilities that decide whether an enterprise agentic marketing capability actually works, from product vision through to the team that runs it. Platform depth and product leadership, held by one lead.
The full-stack quest
Level 1 of 5 · 3/15 capabilities unlocked
Product leadership
Level 1 · Set the vision
Reward: A roadmap marketers want to use
Vision, strategy & roadmap
Own the vision and roadmap for a generative-AI front door that acts as the central UI and orchestration layer for downstream marketing agents.
Journeys & UX for marketers
Design journeys with campaign managers and marketing ops across brief, ideation, audience selection, creative generation, approvals, deployment and post-launch optimisation.
Outcomes into backlog
Translate business outcomes into epics, features and acceptance criteria, then prioritise and sequence the backlog so agile teams ship high-quality increments.
Badge collection
Where it breaks
Agents inherit every weakness in the data beneath them
Four failure points I design out from day one, surfaced by a readiness review before build, with a plan to clear each one.
Data & identity
Fragmented identity
The audience agent targets the wrong person, at machine speed.
Data & identity
Consent that doesn't propagate
Consent captured at the CMP but not carried through Edge to every destination, so opted-out customers still get activated.
Agent reasoning
Real feature, wrong platform
Target On-Device Decisioning is server-side SDK only, yet an LLM will happily recommend it for a Web SDK build.
Ownership
Two agents, one decision
If audience and journey agents both apply eligibility, the rules fork. Every decision needs exactly one owner.
AI doesn’t smooth over a weak foundation. It industrialises it.
Reference architecture
How I architect agentic marketing on Adobe Platforms
One front door for marketers, an orchestrator that enforces human-in-the-loop gates, specialist agents each grounded in one product's slice of truth, and governance living in the tool layer.
Experience layer
One front door for brief, ideation and approvals
Orchestrator
Plans, routes and enforces human-in-the-loop gates
Audience
RT-CDP
Creative
AEM / EDS
Channel
AJO
Compliance
Policy + labels
Insights
CJA + Target
Tool layer
Governed APIs and MCP tools; every claim carries evidence
Foundation
XDM · identity graph · merge policies · consent & governance
Across every layer
Evaluation & observability
Tracing, eval harness, red-teaming, cost and latency SLAs.
Segmentation decides who, the journey decides how. RT-CDP owns eligibility, AJO owns timing and channel, and the orchestrator enforces that split. Never let two agents own the same rule.
Campaign lifecycle
From brief to optimisation, with people at the right gates
Agents accelerate every step of the campaign manager's journey. The approval gate is non-negotiable in regulated industries.
- 01
Brief
Goal, KPI and constraints captured once
- 02
Ideate
Concepts grounded in brand and past results
- 03
Audience
RT-CDP segments with consent applied
- 04
Create
On-brand variants from governed content
- 05
Approve
Human gate: legal, brand and risk review
- 06
Deploy
AJO journeys and paid media activation
- 07
Optimise
CJA lift readout feeds the next brief
Adoption, engagement and impact KPIs on live dashboards
Responsible AI & operations
Governance and product operations, built in from sprint one
Model risk is managed like any other enterprise risk: owned, measured and reported. Cost and quality run as product metrics, so model choices stay explainable to finance and risk.
Responsible AI & model risk
- Governance and model risk management: owners, approvals, model inventory
- Privacy and PII safeguards mapped to the Australian Privacy Act
- Bias and quality testing before and after every release
- Prompt logging and human-in-the-loop review
- Content-policy and brand-safety guardrails
Product operations
- SLAs, incident response, observability and telemetry
- Evaluation harnesses and red-teaming on every release
- Quality: precision, recall, hallucination rate, user satisfaction
- Cost and speed: token usage, caching, RAG, model selection, throughput and latency SLAs
- Continuous improvement loops driven by live signals
The run rhythm after go-live
Agents drift as products, prices, policies and models change. Every agent has a named owner, and every change runs through the eval harness.
Daily
- Quality, cost and latency dashboards
- Alerts routed to named agent owners
Weekly
- Review flagged outputs and approval overrides
- Evals run on every change
Monthly
- Refresh knowledge, offers and rules
- AI cost versus value review
Quarterly
- Red-team and bias testing
- Model and platform upgrade testing
Partnership
One way of working across every function the capability touches
Customers experience the enterprise, not the org chart. An enterprise-grade capability needs all eight functions aligned.
Marketing
Journeys and UX designed with campaign managers
AI/ML engineering
Agent, RAG and evaluation architecture
Data science
Propensity and next-best-action models
Analytics
CJA measurement and impact dashboards
MarTech / AdTech
AEP, AJO, Target and paid media integration
Legal / Compliance
Approval gates and content-policy guardrails
InfoSec
PII safeguards, access control and logging
Risk
Model risk evidence, owned and reported
Change that lasts: I own enablement and training for marketers, and run a feedback loop with power users and a Community of Practice.
How I start
Prove it on one journey, then scale
Phase 1 · Discover
People, data & traps
- Human-centred research with marketers and marketing ops
- Readiness review across data, identity and consent
- Trap register, target architecture and backlog
Phase 2 · Pilot
One use case, end to end
- A high-value campaign on the orchestrator
- Human gates, evals and telemetry live from the start
- Baseline versus lift measured in CJA
Phase 3 · Scale
Operate & expand
- Add agents and channels from the backlog
- SLAs, incident response and cost controls
- Enablement and a Community of Practice
Start where the data is strongest, prove lift on one journey, then scale across channels.
Let’s talk
Looking to take agentic solutions into production?
Please get in touch. I’m always happy to work through what it takes to run agentic marketing and personalisation in production: the data foundations, orchestration, guardrails and measurement that make it work beyond the pilot.
Experience
Leadership and build work
Personalisation Platform Lead
Campfire Digital Services | Current
I lead Campfire’s personalisation platform practice, shaping enterprise personalisation and agentic orchestration across data, decisioning, content and activation. My work is platform agnostic, with deep experience across the Adobe ecosystem and the broader data and technology landscape that powers it.
- Architect enterprise personalisation across customer data, identity, decisioning, orchestration and activation, including Adobe Experience Platform and its wider ecosystem
- Lead solution architecture and technical design across complex enterprise engagements, transformation programmes and RFPs
- Connect marketing platforms with enterprise data products, integration layers and customer intelligence to enable personalisation at scale
- Embed privacy, governance and responsible AI into the architecture from the outset rather than treating them as downstream controls
- Design AI orchestration patterns that coordinate specialised agents across decisioning, content, activation and optimisation
- Define observability and evaluation approaches for AI solutions, including traceability, quality, performance and continuous improvement
- Design AI solutions with production economics in mind, balancing model selection, latency, scalability and cost without compromising quality or control
Enterprise Intelligence & Personalisation
Enterprise Financial Services | Previous
I led personalisation across a complex, regulated financial services ecosystem, connecting marketing technology platforms with data products, decisioning and integration layers to enable intelligent customer orchestration at enterprise scale, with risk, compliance and governance built in by design.
- Led enterprise AI and personalisation strategy across engineering, data, product and marketing teams, translating business problems into scalable technology capabilities
- Designed the architecture connecting enterprise data products, customer intelligence, decisioning and activation platforms to support real time personalisation and orchestration
- Established experimentation and measurement frameworks that connected personalisation and AI initiatives to customer and commercial outcomes
- Led platform modernisation and integration across a complex technology landscape, reducing fragmentation and creating stronger foundations for data driven decisioning
- Built governance into the operating model, ensuring customer data, models, audiences and activation could be managed safely across a highly regulated environment
Trading Decision & Orchestration Platform
Agentic Architecture | Decision Systems | Observability | Production Engineering
I built a trading decision and orchestration platform to explore how intelligent systems can reason over market data, coordinate decisions and turn those decisions into controlled actions. The platform combines model reasoning with deterministic logic, giving me a practical environment to test what works, what fails and what it takes to build agentic systems for production.
Core design principles
- Bring together data, application services, deterministic logic and model reasoning as one decision architecture
- Use specialised agents to reason, validate and refine decisions before action
- Build observability and evaluation into the system so decisions are traceable, measurable and continuously improved
- Keep rules, controls and critical decisions deterministic, using model judgement where reasoning genuinely adds value
- Design for production from the outset with reliability, governance and human oversight built in
Deterministic where possible. Model judgement where it adds value.
Principles
How I design, build and scale intelligent systems
Foundations first
Identity, consent and data quality fixed before agents act, so AI scales value, not errors.
Full-stack depth
AEP, AJO, CJA, AEM and AI engineering held together. The failures live at the boundaries, so I own them.
Builder, not just adviser
I ship agentic systems myself and bring those patterns, evals and guardrails to the teams I work with.
Regulated-industry ready
Shaped in enterprise financial services: risk, compliance and model governance are defaults.
Enablement that sticks
Training, power-user feedback loops and a Community of Practice that marketers own.
Insights
Personalisation, AI, and digital experience thinking
Practical views on modern CX, AI-powered systems, asset management, and how organisations make their digital experiences truly scalable.
Personalisation Is Not Just Decisioning
While AI models, decisioning engines, and real-time data are important, true personalisation requires structured content models, consistent metadata, modular asset design, and clear governance.
Every decision still depends on what can actually be delivered. Even the most advanced AI will surface the same content, just faster, without proper foundations.
The Evolution of Personalisation: From Rules to AI
Personalisation has evolved through four distinct eras: Rules-Based → Data-Driven → Real-Time → AI-Augmented. Each era brought new capabilities and new complexity.
Technology accelerated through each era, from simple rules to real-time events to adaptive systems. But operating models struggled to keep up. Personalisation evolved faster than its supporting systems.
The Shift That Actually Matters
The real transformation isn't in capability, it's in readiness. Design content for scale, treat DAM as infrastructure, align content/data/journeys into one system, embed governance from the start.
The industry has invested in decisioning engines, real-time infrastructure, and AI. But the real shift is in readiness: designing systems that hold together at scale.
Personalisation as an Operating Model
Successful organisations don't treat personalisation as a one-off initiative. They treat it as a system: Start small → Measure → Learn → Adapt → Scale.
Not everything needs to be real-time. Not everything needs AI. What matters is clear use cases, practical execution, and strong foundations. Scaling complexity without readiness only amplifies the problem.
Contact
Let’s build what’s next, together
Open to conversations about agentic marketing, enterprise personalisation and responsible AI: product leadership, architecture and hands-on delivery with strong operational rigour.