ISYS90026 · Concepts In Information Systems
AI Transformation in Business and Exam Synthesis
This final ISYS90026 topic does two jobs. From the Datacom guest lecture it teaches what organisations must get right to make AI pay off: start with the problem, not the technology, and cross the gap from adopting (bolting tools onto old workflows) to adapting (redesigning workflows around AI). From the wrap-up it turns the whole subject into exam technique — the synthesis question hands you an unseen AI case and rewards stitching three or more of the four themes together rather than reciting one. Expect AI-vocabulary MCQs plus a cross-theme essay on the closed-book, 50% hurdle exam.
What this chapter covers
- 011. Adopting vs adapting — tools bolted on (minor gains) vs workflows redesigned around AI (real transformation); 78% use gen AI but only 21% redesign a workflow
- 022. Start with the problem, not the technology — why 95% of enterprise AI projects fail and most firms stay stuck exploring
- 033. Three Horizons of AI-enabled work (IFTF) — H1 powerful but limited, H2 rapid advance (agentic, domain models), H3 fundamental transformation
- 044. Five leadership AI mindsets — unaware, paralysed, checkbox, false summit, and the goal: strategic
- 055. The Frontier AI organisation (Harvard) — Foundations → Modern AI-driven IT → back office → front office, on the Process/People/Technology pillars
- 066. How AI threads through the four themes — core competence, competitive advantage (Porter & Millar), alignment, governance/sourcing, change, responsible business
- 077. Where value shows up — AI-first/agentic customer experience, zero-click search, and the labour-to-compute cost shift; Datacom's five lessons
- 088. Exam synthesis — the 10 MCQ + 4 essay structure, What/Why/How, and connecting three or more themes to one unseen case
Cross-theme synthesis on an unseen AI-first case
- +3Adopt vs adapt — diagnose the mindset. Meridian has adopted (a copilot) but not adapted (workflows unchanged, no earnings impact) — the classic 78%-use / 21%-redesign gap. Its leadership is in the checkbox / false-summit mindset, with the board treating AI as "buy it now". The fix is the strategic mindset: a deliberate, outcome-linked AI strategy — i.e. business-IT alignment, with IT capability reshaping strategy (the competitive-potential perspective).
- +3Foundations and governance. Apply the Frontier-AI blueprint bottom-up: Meridian has skipped Foundations — Process (no AI charter or strategy linked to outcomes), People (no literacy/adoption plan) and Technology (scattered, dirty data). Map to IT governance: decide AI principles, architecture and investment, and pick a Federal / IT-duopoly archetype — an AI Council or Centre of Enablement so business and IT co-own outcomes ("leadership is collective").
- +3Sourcing. Split the capability: the proprietary credit/fraud model is strategic and needs a rare domain-plus-IT mix, so build/insource it and keep the IP and data in-house as a potential core competence; commodity AI (document OCR, generic chat) goes to cloud/outsource, leveraging a partner ecosystem with a sovereign-data option appropriate for a regulated bank.
- +6Change management and responsible business. Use Kotter: establish urgency (the neobank is taking customers; AI is "electricity-scale"), form a collective guiding coalition ("AI can't be one person's job"), deliver short-term wins, and anchor via a living strategy ("the bar keeps moving"). Embed responsible-AI guardrails — fairness in automated lending, privacy, transparency and compute/Green-IT cost — framed as strategic CSR, not a compliance afterthought.
Key terms
- Adopting vs adapting
- Adopting means bolting AI tools onto existing work (a copilot rollout, a pilot) for minor gains; adapting means redesigning the workflow around AI so the work itself changes shape. The gap between the two defines AI maturity — 78% of firms use gen AI but only ~21% have redesigned a workflow.
- Three Horizons of AI (IFTF)
- A foresight frame: H1 = AI is powerful but limited (truth/verification gaps, hallucinations, weak ingenuity); H2 = rapid advance (specialised domain models, multi-model ecosystems, agentic and physical/robotic AI); H3 = fundamental transformation of industries, organisations, services and how work is organised.
- Leadership AI mindsets
- Five postures a leader can take toward AI — unaware ("it's just an IT issue"), paralysed, checkbox ("do something to look current"), false summit ("we rolled out Copilot to 30 people"), and the goal, strategic ("ask the hard questions that drive real transformation").
- Frontier AI organisation
- Harvard's term for a new class of organisation that puts AI at the heart of its business strategy to reinvent how it operates, innovates and augments human capacity at scale. Its blueprint stacks Foundations, Modern AI-driven IT, back office and front office.
- Foundation pillars (Process / People / Technology)
- The three pillars under the Frontier-AI stack: Process (AI charter, governance/risk/ethics, regulatory alignment, outcome-linked strategy), People (AI literacy, culture/adoption, safe tooling, sandboxes) and Technology (clean-data platform, agent platform, security/observability).
- Agentic / MCP interface
- An AI-mediated experience where agents act across systems via automation APIs (such as the Model Context Protocol), increasingly generating the user interface from a customer's objective rather than serving a fixed website.
- Zero-click search
- A customer-experience shift where AI answers a query directly (e.g. at the top of a results page or via a voice assistant) so the user never visits the source site — a buyer-power and substitution shock for businesses that relied on web traffic.
- Labour-to-compute cost shift
- The structural change as routine work is automated and service costs collapse: spend moves from labour to compute, giving AI-first firms a compounding cost advantage at near-zero marginal cost.
- AI Centre of Enablement (CoE)
- An operating-model structure (alongside an AI Council or Lab) that turns experimentation into organisation-wide capability — Datacom's lesson that "federated doesn't scale" and strategy needs an operating model to come to life.
AI Transformation in Business and Exam Synthesis FAQ
What is the difference between adopting and adapting AI?
Adopting means bolting AI tools onto existing work — rolling out a copilot or running a pilot — which yields only minor productivity gains. Adapting means redesigning the workflow around AI so the work itself changes. The exam-critical statistic is the gap: about 78% of organisations use gen AI but only ~21% have actually redesigned a workflow, which is why most see no material earnings impact.
Why do most enterprise AI projects fail?
Because firms treat AI as a technology rollout rather than a strategy, change and governance problem — "enabled isn't enough". They start with the tech instead of a problem to solve, skip the foundations (charter, literacy, clean data), and never move leadership from a checkbox or false-summit mindset to a strategic one. Around 95% of enterprise AI projects fail to deliver.
What is the synthesis question and how do I answer it?
It is the wrap-up essay that hands you an unseen AI case and asks you to advise using frameworks from across the subject. The marks come from connecting three or more themes — core competence/competitive advantage, alignment, governance/sourcing, change management, responsible business — to the same case. Lead with the framework name, map it to the case evidence, recommend, then move to the next theme.
How is the ISYS90026 final exam structured?
It is a closed-book, invigilated exam taken through Respondus LockDown Browser, worth 100 marks (50% of the subject): 10 multiple-choice questions plus four essay sections, one per theme. There is no penalty for wrong MCQs, partial credit is available on essays, and it is a hurdle — you must score at least 50/100 to pass the subject.
Do I need to memorise a new AI framework for this topic?
No. The AI material is a lens that re-reads the earlier frameworks rather than a new model to memorise. You should know the vocabulary (adopt vs adapt, the Three Horizons, the five mindsets, the Frontier-AI blueprint and the Process/People/Technology pillars) and be able to connect each course theme to an AI case.
Is this page affiliated with the University of Melbourne?
No. This is an independent AskSia study resource for students taking ISYS90026; it is not produced or endorsed by the University of Melbourne, Datacom or any guest lecturer. Always confirm assessment details against the official Canvas subject page and current handbook.
Exam move
Treat this topic as two layers. First lock the AI vocabulary the MCQs test: adopting versus adapting (the 78% / 21% gap), the Three Horizons, the five leadership mindsets (especially the false summit, where a copilot rollout is mistaken for transformation), and the Frontier-AI blueprint resting on the Process / People / Technology pillars. Then drill the synthesis essay, which is the real differentiator: practise taking an unseen AI case as a central node and spoking out to at least three themes — diagnose adoption-not-transformation (alignment), locate the missing foundation layer and pick a governance archetype, split sourcing so the strategic model is built in-house while commodity AI is bought, lead the change with Kotter, and add responsible-AI guardrails. Keep problem, root cause and evidence separate, use the relevant lenses rather than every lens, and remember the exam is a hurdle so breadth across all four themes beats deep mastery of one.
