The question every board, CEO and CIO must answer in 2026 isn’t whether to use AI. It’s whether to use AI to improve what you have, or to start again. Most organizations are getting this choice wrong, defaulting to whichever option matches their risk appetite, rather than applying clear strategic criteria.
Rewiring treats existing processes, teams and systems as the frame, using AI as the wiring that makes them faster and smarter. The enterprise stays recognisable. Org charts shift modestly. Underneath, AI accelerates throughput and cuts manual effort e.g. AI co-pilots in legal review, ML-driven demand forecasting, generative AI auto-resolving Tier 1 support tickets.
Rebuilding treats the current operating model as legacy and uses AI as the architectural foundation for something structurally different. Entire functions may disappear or be reborn. Processes are redesigned from first principles with AI at the core, not bolted on. Think: a digital-only insurance carrier built around AI underwriting by default, not as an add-on.
Neither is universally correct. The organizations winning this decade are applying disciplined criteria and increasingly, sequencing both.
The decision framework
Five questions determine the right path, and they need to be asked together, not in isolation.
Is the operating model the constraint, or is execution? If processes are sound but slow and error-prone: rewire, AI removes friction without touching the underlying logic. If the architecture itself is fragmented and siloed by design, rebuild. AI plugged into a broken process just produces faster, better-documented brokenness.
How much runway do you have? Rewiring delivers ROI in 3–12 months. Rebuilding takes 18–48 months before material value shows up. If competitive pressure demands proof of AI value within a year, rewire first. If an AI-native competitor has already entered your market with a structurally lower cost base, incremental improvement won’t close that gap, only rebuilding will.
Can your people absorb the change? A workforce that’s risk-averse or change-fatigued can adopt AI-in-place without existential threat to most roles. A rebuild without genuine leadership mandate and a credible workforce transition plan isn’t transformation, it’s poorly managed redundancy with better PR.
How bad is the technology debt, really? Most AI use cases can be delivered via APIs and abstraction layers without core system replacement. Rebuild only when the estate is so fragmented that a unified data layer or real-time decisioning is structurally impossible otherwise.
Does the prize justify the disruption? Bounded efficiency gains of 10–25% rarely justify a rebuild’s cost and risk. Step-changes in unit economics or customer proposition do.
Who should decide
This is a capital allocation and talent strategy decision with technology implications, not a technology decision. The most common governance failure is letting the CIO or a transformation consultancy own it unilaterally.
The decision table needs the CEO, who owns the risk-return trade-off and the mandate to change; the CFO, who must model the economics of both paths honestly, including the productivity dip during transition, not just peak-state ROI; the CHRO, who needs a credible transition strategy in place before the decision is taken, not after; the CIO, who assesses technical feasibility but shouldn’t be making the strategic call alone; and business unit leaders, whose operational insight and buy-in are non-negotiable. An AI-literate independent board voice helps prevent both excessive caution and hype-driven overreach.
Costs, benefits and where maximum value sits
Rewiring’s ceiling is real, gains are bounded by the existing model, and it risks “AI-washing”: surface deployment without structural impact. But it’s fast, lower risk, preserves institutional knowledge and compounds across multiple waves over several years.
Rebuilding can deliver 30–60% structural cost reduction and capabilities simply unavailable to a rewired legacy model, but it carries a real failure rate (high for large transformations), heavy upfront investment and a multi-year J-curve before returns appear.
Maximum value rarely comes from choosing one exclusively. It comes from sequencing: rewire to generate cash, capability and credibility, then rebuild the two or three domains where AI-native architecture creates a genuine moat, while continuing to rewire everything else.
Case in point: An Australian tourism and cruise operator
Consider one of Australia’s largest integrated tourism and cruise businesses, simultaneously a B2C retailer, a B2B distributor to thousands of agency and wholesale clients globally, an aggregator marketplace for 1,800-plus independent tourism operators, and a cruise operator with offshore shared services spanning finance, customer contact and content management.
By 2024, the pressures had converged: AI-native travel platforms eroding acquisition economics, independent operators demanding dynamic pricing the platform couldn’t offer, and offshore cost structures under threat from automation. Leadership’s assessment found a split picture. The B2C and shared-services functions were sound but manual, a rewiring opportunity. The aggregator marketplace’s static catalogue and rules-based search were the actual constraint, no amount of AI on top would fix that. It needed rebuilding.
Rather than choose one path, the executive team sequenced three horizons. Horizon 1 rewired customer contact (AI triage cut Tier 1 escalations by 34%), content management (AI drafting cut operator listing time by 70%, eliminating a 23-day onboarding backlog), finance operations, B2C personalization (higher email revenue) and cruise crew scheduling (15% lower overtime). Within 18 months this delivered a million in annualised savings, funding and validating the next move.
Horizon 2 rebuilt the marketplace itself: AI-native semantic search lifted booking conversion by 24%; opt-in dynamic pricing lifted operator revenue per booking 16% for the first cohort; automated onboarding cut new-operator time-to-live from 23 days to three.
Critically, the offshore teams whose roles were most exposed to automation weren’t reduced, they were redeployed into quality assurance and operator onboarding, work that leveraged the institutional knowledge AI couldn’t replicate. Zero redundancies came out of Horizon 1. That decision wasn’t only ethical; the content quality gains from experienced specialists focusing on QA rather than production were measurable.
The lesson generalises well beyond travel: rewiring generated the cash, capability and credibility that made rebuilding possible. Neither path alone would have delivered the same outcome, and the sequencing mattered as much as the technology choices themselves.
What this means for CIOs
Start with rewiring, generate tangible ROI within 12 months and use it to build capability and board trust. Watch for your structural ceiling: the point where further rewiring yields diminishing returns because the model itself is the constraint. That’s your signal to rebuild selectively. Don’t rebuild everything; identify the two or three domains where AI-native architecture creates real competitive advantage and rewire the rest. And treat workforce transition as a strategic priority from day one, not an HR afterthought bolted on after the technology decisions are made.
The rewire-or-rebuild question isn’t a technology question. It’s a question about what kind of enterprise you’re choosing to become. The CIOs who get this right won’t be the ones who pick a side, they’ll be the ones who know exactly when to switch.
This article is published as part of the Foundry Expert Contributor Network.
Want to join?