AI Enabling the Modern IT Operating Model
0 likes25 views
Integrating AI capabilities into the Operating Model concepts I've been working with for 2 decades! Features AI augmentation, Inference, AI Insights Engine and Data Lakehouse concepts.
AI Enabling the Modern IT Operating Model
- 1. AI, Your New“Partner”in Operating Model Uplift COSAC APAC 2025 David Favelle [email protected]
- 2. Our 50-Year IT Struggle Image Credit: The New Yorker 2021 • Decades wrestling with frameworks, standards, & capability uplift • SABSA, COBIT, ITIL, IT4IT, TOGAF, - all necessary, and exhausting • Add leadership changes, technology disruption, bad sourcing changes, M&A, … • How far have we come…really? And then along comes AI
- 3. Did AI Arrive for you in 2023/24? 2025: Year of the Agents/Co-pilots “I thought we’d have more time”
- 4. Meanwhile in the USA… Chat GPT Gov “ChatGPT Gov includes access to many of the same features and capabilities of ChatGPT Enterprise, such as: Saving and sharing conversations within their government workspace, and uploading text and image files. GPT-4o, our flagship model, excelling in text interpretation, summarization, coding, image interpretation, and mathematics. Custom GPTs that employees can build and share within their government workspace. An administrative console for CIOs and IT teams to manage users, groups, Custom GPTs, single sign-on (SSO), and more." https://openai.com/global-affairs/introducing-chatgpt-gov/?utm_source=tldrai
- 6. SalesForce = Agentforce for 2025
- 7. ServiceNow lifts the lid a little more…
- 8. Agentic benefits Constrains the work to the platform and value stream becomes more observable and manageable Improved data on our human and AI effort in supporting of digital products and services Reduces error rates in repeatable tasks, while remaining non-deterministic Frees up valuable human effort for working“on the system”rather than in it “AI agents will become an integral part of our daily lives, helping us with everything from scheduling appointments to managing our finances. They will make our lives more convenient and efficient.” – Andrew Ng, Co- founder of Google Brain and Coursera.14 June 2024“
- 9. Battle of the Tech Stacks “Apples not falling far from trees”or“dogs looking like their owners”?
- 10. 2025 may go like this… 1. Lines of business will push hard with AI to maintain competitive parity or maybe get some incremental advantage • Best of breed, local scenarios 2. Corporate will look at ways to leverage AI to reduce cost and drive performance • On-platform agents in ERPs 3. IT will transition to AI coding and ITSM/Ops will scramble to recalibrate • Agents starting to do some heavy lifting 4. Suppliers will be leveraging AI to take out cost and stay within their contractual obligations • Agents, automation 5. Security and governance folks will react to AI related changes in the regulatory environment and try to close the gaps • Redefining risks and controls The heat is on cycle times accelerating.
- 11. But, we didn’t ban cars, we adopted: • speed limits, • safety standards, • licensing requirements, • drunk-driving laws & other rules of the road.” Bill Gates How should we respond to AI? “Soon after the first automobiles were on the road, there was the first car crash… Eyes on the opportunity & the guardrails
- 12. Brief History of IT Capability Uplift • Isolated capabilities built to address emergent & painful gaps e.g. Service Desk, Change Control, DR • No overarching “systems model” to build toward • Rudimentary, isolated outcome metrics • Monolithic control frameworks, generic standards, and “arguable” best practices • Capability was built at the hotspots of risk, pain and cost • Improvements undertaken on a best-efforts basis and mostly “getting away with it” and sometimes not How might we harness AI to bootstrap capability?
- 13. Is the answer to AI success about the data? • No. We’ll collapse trying to get the data right • It’s more about inference than analytics • ML is incrementalism • Agents are arbitrage • Pattern matching incidents etc is just “efficient” • Inference is about: • Working with scant or partial data • Qualitative insights based on best practices and targeted metrics • This is what effective consultants do, especially at low maturity levels and high urgency scenarios
- 14. Inference-Led Role Augmentation • A personal, well-informed analyst • Reasoner agent, ChatGPT5? • Ability to handle structured and unstructured data • Upload in real time along with well structured prompts • Ability to advise across the IT Op Model including generic and industry vertical common practices • ITIL, COBIT, SABSA • You don’t have to train it on generic IT principles and it can react very well to local information input as part of the prompt
- 15. How to Outpace the change? 1. Define the Operating Model requirements (see next slides) 2. Evaluate & Define the AI Roles: ▪ Think: Advisor/Analyst Augmentation ▪ Harness: Agentic Augmentation ▪ Enact: Automation & ML 3. Create a plan including defined value, risk and performance outcomes 4. Create an AI Insights Engine 5. Drive short cycles of high value improvements
- 16. IT Operating Model IT Strategy, EA & Governance Business Strategy Technology & Sourcing Roadmaps Service Portfolio (target), Medibank Assets & Sourced Services Traditional (Aspirational) IT Operating Model 16 Service Portfolio Internal & Partner Sourced Services Automated Workflows, Controls & Reporting Value Streams & Capabilities (processes) Measurement & Improvement Service Brokerage & Partner Integration Organisational Design Leadership & Culture Engagement Model & Demand The IT Operating Model concept has emerged as a way of fully expressing the integrated set of capabilities needed to run modern IT It is a multi- dimensional system operating together to deliver business technology value
- 17. Example “As-Built” Full Capability Operating Model 3rd Party Services PLAN (Strategy to Portfolio) BUILD (Requirements to Deployment) DELIVER (Request to Fulfil) RUN (Detect to Correct) System Integration providers Infrastructure & Operations Services Supporting Capabilities Management & Control Capabilities ENGAGE (Manage Relationships) Manage Partners Integrate Partners Customer Engagement Demand Gathering & Shaping Idea/Concept Validation Portal Management Content & Communications Experience Reporting & Improvement Service Desk Asset Mgmt Knowledge Mgmt Configuration Mgmt Resource Mgmt Reporting Workforce Management Financial Management Strategy & Planning Continuous Improvement ICT Governance & Risk Enterprise Architecture Operations Assurance Disaster Recovery Tools Integration Change Control Fulfilment Assurance Performance Mgmt Vendor Management Delivery Control Analytics Mgmt Demand Management Portfolio Management Service Design Platform Lifecycle Management Service Lifecycle Management Delivery Management Requirements Management Development Management Platforms & Tooling Test Management Transition Management Service Monitoring Incident Management Change Management Event Management Problem Management Operations Management Offering Management Consumption Control Service Recharge Request Fulfilment Catalogue Management Service Automation & Orchestration Cyber Security Security Solutioning Policy Management Security Operations Security Architecture Design Partner Security Assurance Value Creation Value Delivery ENGAGE (Manage Experiences) Corporate Lines of Business End Users Customers AI & ML Management Information Management
- 18. IT Strategy, EA & Governance Business Strategy Portfolio, technology & Sourcing Roadmaps Value Streams & Capabilities Measurement & AI Optimisation Service Brokerage & Partner Integration Organisational Design, AI Role Augmentation Leadership & Culture Engagement Model & Demand Security, Risk & Governance ML, Automation, IT Mgmt Platforms Product & Service Portfolio AI Insights Engine DRIVERS OPERATING MODEL COMPONENTS Lakehouse AI Advisor Introducing: The AI Enabled IT Operating Model
- 19. AI Insights Engine (AIIE): - Balancing Flow, Risk & Cost of Delay Unified Visibility • Integrates data from Dev, ITSM and Ops tools e.g. ServiceNow, Jira, Splunk, and other key systems • Breaks down data silos for a comprehensive IT operating view Key Metrics & Insights • Cost of Delay (CoD): Quantifies the economic impact of delayed initiatives • Operational Risk: Predicts incident likelihood and compliance exposure • Flow Metrics: Monitors work-in-progress, queue times, and flow efficiency based on Reinertsen’s principles AI-Driven Recommendations • Provides actionable insights for weekly prioritization and monthly performance reviews • Balances speed (feature delivery) with stability (risk and compliance) Outcome • Empowers the CIO with a single source of truth for data- driven decision-making • Enables proactive, optimized IT capability improvements
- 20. AIIE: Proof of Concept & Roadmap Proof of Concept (PoC) Focus Use Cases: • Prioritization: Ranking backlog items using CoD and risk scoring • Risk Monitoring: Predicting incidents tied to upcoming releases Data Sources: • Initial integrations with Dev, ITSM and Ops tools ServiceNow, Splunk (expandable as needed) across a limited product portfolio Architecture Overview Data Ingestion: ETL pipelines to collect data from multiple sources Analytics & AI Layer: • Descriptive & predictive models (incidents, CoD calculations) • Basic recommendation engine for decision support CIO Dashboard: • Intuitive, real-time visualizations for KPIs and “what-if”scenario analysis Resource & Skill Requirements • Cross-functional team: Solution Architect, Data Engineer, Data Scientist/ML Engineer, Front-End Developer, and Product Owner • Timeline: 4–8 weeks for a robust PoC with iterative enhancements Next Steps & Expansion • Integrate additional data sources & refine ML models • Extend use cases to include compliance, resource allocation, and vendor management
- 21. Service Portfolio & Service Models 1. Configuration Drivers Internal & Partner Delivered Services Business/IT Strategy & Business Model Attributes AI in Operating Model Transformation 21 3.Designing Capabilities Jobs & Roles, AI Aug Process Definition AI & Automation Requirements Scorecards& Reports Skills & Behaviours Team scope & structure 2.Aligning Dimensions IT Operating Model Dimensions Engagement & Demand Service Brokerage & Integration Organisation Design, Ai Aug AI & Automation Architecture Measurement & AI Optimisation Lifecycle/ ValueStreams 4. Realising the Change Job Descriptions AI training Process Manual Service Agreements Supplier Requirements Staff/Team Development Plans Team/Function Definition AI & Tool Functions Service Performance Reports Portfolio & Attributes A lot of artefacts need to be generated, but they are not that hard using AI
- 22. Options to get this done? 1. Engage a big consultancy firm! 2. Spend all you have on ServiceNow 3. Outsource your IT and hope that the MSP has what you need 4. Delegate it to your team! 5. Send everyone to Six Sigma training and hope? 6. Build your own GPT and get started! …wait, what?
- 23. Hmm… Sounds complicated But there are a few ways: Have any of you done this? Let’s have a look: https://chatgpt.com/canvas/shared/67ab2537d4d08191a56df96227385bef 1. Build your own LLM and train it with your content & hardware behind a firewall 2. Build one from as-a-service components: • Salesforce, OpenAI, Azure, Google, AWS… 3. Go to market for a solution 4. Build a GPT on OpenAI
- 24. AI Use Cases for Operating Model Capability Uplift • Op model“system advisor” • Analysis of business attributes • Design Op Model components • AI Insights Engine design & build • Current State Capability analysis • Current State Service performance analysis • Artefact creation • Tooling design and coding • Op model“system advisor” • AI Insights Engine • Continuous • Service Improvement • Capability Improvement • Artefact Refinement • Tooling maintenance and refinement Build Run
- 25. Some Pros and Cons • Always on = 24 x 7 • Immediacy of responses • Not in meetings • Non emotive • Data driven • Multi modal • Available from many sources • Price is dropping, value is increasing • Can scale up and down easily • Hallucinations • Rapid release cycles • Architecture still evolving • Access controls/limits • Not real world aware • Memory not persistent - yet • Deception (on sub-goals) • Agent sprawl • Potential chaos if poorly managed Positives Challenges
- 26. Potential next steps for you 1. Assess your business’s AI strategy • There will already be a declared intent 2. Assess the skillset and mindset of your team • Tailwinds or headwinds? 3. Decide where you can make an impact with AI • Impact assessment within your span of control, or • Engage in the broader AI effort 4. Build or buy your GPT capability and Op Model insights engine • Maybe there’s a PoC you can do to engage folks 5. Start small, deliver value early and often • Get “match fit” by making AI part of your daily workflow • Keep working on your domain expertise, it still matters https://www.cio.com/article/3815112/25-enterprise-tech-predictions-and-goals-for-2025-from-apac- cios.html?utm_date=20250206015456&utm_campaign=CIO Australia First Look&utm_content=slotno-7- readmore-We asked 25 thought leaders in the Asia-Pacific region to share their digital transformation predictions and goals for the year.&utm_term=ANZ&utm_medium=email&utm_source=Adestra&huid=7ba6f364-7799-4d83- 9d90-0dc2a19436aa
- 27. Key Take Aways 1. Agents and Inference will be a big deal in 2025 2. Harnessing AI is the key to getting in front of it 3. The quality of your prompts will determine the AI output, invest some time 4. Experiment with low cost LLMs. It will get you a long way 5. AI Insight Engine can underpin data driven decision making and get your IT in control of the Operating Model as an Integrated system
- 28. Topics for drinks? • Agentic & Co-pilot releases • ServiceNow March ‘25 • OpenAI Vs Google Vs… • Guardrails: • Aus Fed gov (DTA) • Global (NIST, ISO 42001) • Digital Twin emergence • Product & Service • Operating Model Capability • NVIDA NIM & Blueprints