Agile Product Delivery with ITIL 5 and AI

For years, Agile and ITSM lived in an uneasy truce. Agile teams moved fast, embraced change, and shipped in two-week sprints. ITSM frameworks enforced change advisory boards, ticket queues, and release windows measured in months. The tension was real — and it slowed organisations down.

ITIL 5 breaks that truce — not by picking a side, but by reframing the conversation entirely.

The shift: from service management to product management

ITIL 4 positioned itself as a practices-based framework compatible with Agile and DevOps. ITIL 5 goes further: it replaces the concept of service management with product and service management as the primary lens.

This is not semantics. It changes how teams are structured, how work is prioritised, and how value is measured.

For agile delivery teams, the ITIL 5 product-centric model is the closest any ITSM framework has ever come to speaking the same language.

The PSLM — an agile delivery loop you already know

The Product & Service Lifecycle Model (PSLM) is ITIL 5's replacement for the Service Value Chain. Its 8 phases map directly onto how modern product teams already work:

Discover → Design → Acquire → Build → Transition → Operate → Deliver → Support

Here is how each phase connects to agile practice:

The key insight: the PSLM is not a waterfall. In ITIL 5, phases can run in parallel and loop back. A Build sprint can feed directly back into Design — exactly like agile iterations.

Where AI changes the delivery model

ITIL 5 introduces the AI Capability Model (6C) — six distinct AI capabilities that product and service teams can apply across the PSLM. This is not a generic "use AI" recommendation. Each capability maps to specific delivery activities:

Creation (C1) — Accelerate build and documentation

AI generates code scaffolding, test cases, runbooks, release notes and architecture diagrams. What used to take a day takes an hour.

Apply at: Build, Transition, Support phases.

Curation (C2) — Surface the right information at the right time

AI filters and ranks signals from monitoring tools, change logs, incident histories, and customer feedback. No more alert fatigue — only contextual, prioritised information.

Apply at: Operate, Support, Discover phases.

Clarification (C3) — Understand intent from natural language

AI interprets natural language requirements, ticket descriptions, and user stories — reducing ambiguity between product owners and delivery teams, and between users and the service desk.

Apply at: Discover, Design, Support phases.

Cognition (C4) — Augment decisions with reasoning

AI models analyse patterns across incidents, changes, and deployments to predict impact, recommend approvals, and flag risks. Change Enablement becomes data-driven rather than opinion-driven.

Apply at: Design, Transition, Operate phases.

Communication (C5) — Automate routine interactions

Conversational AI handles first-line support, status updates, onboarding flows, and knowledge retrieval — freeing human teams for complex problem-solving.

Apply at: Deliver, Support phases.

Coordination (C6) — Orchestrate multi-agent workflows

Multi-agent AI systems coordinate across monitoring, incident management, change enablement, and deployment pipelines — running autonomous remediation workflows with human approval gates where needed.

Apply at: Operate, Transition, Support phases.

The 5 ITIL 5 practices that make agile delivery work at scale

Adopting the PSLM and the 6C model without operational discipline is just fast chaos. These five ITIL 5 practices are the ones agile teams most frequently under-invest in — and that create the most problems at scale.

1. Change Enablement (not Change Management)

ITIL 5 reframes change management as Change Enablement — the goal is to accelerate safe change, not to create control theatre. For agile teams this means:

• Standard changes for routine deployments (no approval required, automated pipeline)
• Normal changes for moderate-risk releases (lightweight peer review, not a CAB)
• Emergency changes with a fast-track path that doesn't bypass post-implementation review

AI (C4 — Cognition) classifies change risk automatically, reducing the manual overhead of categorisation.

2. Deployment Management

ITIL 5 distinguishes Deployment Management (moving software from build to production) from Release Management (managing the availability of a product to users). Agile teams often conflate the two, creating release confusion.

Separate your deployment pipeline from your release cadence. Deploy continuously; release deliberately.

3. Incident Management with AI triage

At scale, incident volume overwhelms human triage. ITIL 5 supports an AI-first triage model where:

• C2 (Curation) filters noise and surfaces P1/P2 incidents
• C4 (Cognition) correlates incidents with recent changes or known problems
• C6 (Coordination) triggers autonomous remediation scripts for known error patterns
• Humans focus on novel incidents and post-incident reviews

Mean Time To Detect (MTTD) and Mean Time To Restore (MTTR) improve without adding headcount.

4. Product & Service Relationship Management (new in ITIL 5)

The ITIL Service Relationship Model (new in ITIL 5) formalises how value is co-created between provider and consumer across the lifecycle. For agile teams, this translates to:

• Defining explicit value outcomes at the start of each product cycle (not just feature lists)
• Involving consumers in Design and Discover phases, not just Deliver and Support
• Measuring relationship health alongside technical metrics (NPS, adoption rate, STP rate)

5. Continual Improvement embedded in the PSLM

In ITIL 4, Continual Improvement was a separate practice bolted onto the SVS. In ITIL 5, improvement is built into every PSLM phase. Agile retrospectives become ITIL 5 improvement cycles — feeding directly into the next Discover or Design phase.

Track improvements in an Improvement Register with owners, target dates, and measurable outcomes.

Metrics — bridging DORA and ITIL 5

Agile delivery teams often measure success with DORA metrics. ITIL 5 governance uses KPIs. Here is how to unify them:

Track both sets of metrics. DORA tells you how fast your delivery pipeline moves. ITIL 5 KPIs tell you whether that speed is creating or destroying value for consumers.

A practical starting point

If you are an agile delivery team looking to adopt ITIL 5 practices without turning into a process-heavy ITSM shop, start here:

1. Map your product to the PSLM — identify which phases you have good coverage of and which ones are informal or missing entirely (usually Acquire and Transition).
2. Define your change model — separate standard, normal and emergency changes. Automate the standard path.
3. Pick two 6C capabilities — start with C2 (alert curation) and C5 (first-line AI support). These deliver fast ROI with low organisational risk.
4. Run a joint retrospective with your ITSM team — align on a shared improvement register with outcomes, not just action items.
5. Use the ITIL 5 Service Design Builder to document the full product blueprint — it will surface gaps in your team structure, supplier dependencies, and financial model before they become production problems.

Further reading

• ITIL 5 — What's Changing and What It Means for Your ITSM Practice
• ITIL 4 — Complete Guide
• ITSM vs ESM — What's the Difference?