{"id":34446,"date":"2026-06-02T10:54:00","date_gmt":"2026-06-02T08:54:00","guid":{"rendered":"https:\/\/askme.it\/insights\/ai-is-changing-how-it-works-five-moves-that-make-the-difference\/"},"modified":"2026-03-26T12:22:56","modified_gmt":"2026-03-26T11:22:56","slug":"ai-is-changing-how-it-works-five-moves-that-make-the-difference","status":"publish","type":"insights","link":"https:\/\/askme.it\/en\/insights\/ai-is-changing-how-it-works-five-moves-that-make-the-difference\/","title":{"rendered":"AI is changing how IT works: five moves that make the difference"},"content":{"rendered":"
\n

Before 2023, AI in the enterprise was managed by centralized data science teams. Today it enters organizations through three distinct channels: commercial software with embedded AI (such as Microsoft 365 Copilot or Salesforce Einstein), specialized solutions adopted independently by business units for specific needs, and foundational models developed or customized internally by central IT teams.<\/p>\n

This fragmentation is good news \u2014 it brings use cases closer to those who understand operational problems \u2014 and a problem at the same time. Who coordinates everything securely and efficiently? How do you prevent uncoordinated local decisions from creating systemic risks or resource waste?<\/p>\n

From conversations with over 200 CIOs, a recurring pattern emerges: the organizations that get the most value from AI are not those that find the most innovative use case. They are those that make the structural changes needed to support distributed innovation in a coherent way.<\/p>\n

An executive AI council as a starting point<\/h2>\n

Few technology waves in the past have required the involvement of the entire leadership team. AI does, because its implications touch every business function. Verizon structured a cross-functional AI Council where each executive brings the expertise of their area to build a cohesive AI strategy. Participants include the global CIO, Chief Data Officer, Chief CX Officer, CISO, General Counsel, HR, financial controlling, and marketing and business development teams.<\/p>\n

The council meets with structured agendas, minutes, and formal follow-ups \u2014 like a board meeting \u2014 and covers topics ranging from AI adoption and skills development to responsible governance, to the infrastructure needed to experiment and scale. Executive managers then relay discussions to their operational areas, ensuring alignment between strategy and execution.<\/p>\n

Center of Excellence and Community of Practice: two different tools for two different problems<\/h2>\n

Verizon’s GenAI Center of Excellence (COE) has 15-20 experts who dedicate between 30% and 50% of their time to the structure, with skills spanning AI scientists and engineers, enterprise architecture, responsible AI, platforms, application development, risk management, legal\/privacy, compliance, and communications. The COE operates as an interface between distributed teams developing AI use cases and central platform teams, ensuring GenAI execution is aligned with enterprise strategy, building standard design patterns, and developing reusable services with embedded controls.<\/p>\n

Vizient, a healthcare services provider, chose a complementary approach with a GenAI Community of Practice (CoP). A CoP is a less formal, voluntary structure that involves individuals interested in a domain without necessarily being dedicated experts. It is not “governed” by IT, even though IT staff participate. Vizient’s CoP includes AI application specialists from the central team and “champions” for specific employee segments \u2014 software engineers, data analysts, sales and services staff \u2014 who work with a digital employee experience researcher to map AI’s implications on existing roles and redefine human-machine relationships.<\/p>\n

Measuring value, not just costs<\/h2>\n

Distributed teams tend to fall in love with a use case or a demo without evaluating the economic implications. Development, integration, deployment, and operation costs for GenAI solutions can be significant at scale, generating what CIOs call “value leakage.”<\/p>\n

Verizon built an integrated FinOps curriculum for GenAI, designed for all leaders and employees, that teaches how to balance LLM costs and value, how to choose among different deployment and hosting approaches, and how to optimize token consumption through techniques such as prompt tuning, caching, and reuse. A concrete observation from experience: at least 50% of searches on GenAI chat interfaces are duplicates. Semantic caching \u2014 which recognizes when two prompts convey the same meaning and returns the already-computed response \u2014 significantly reduces API costs without impacting quality.<\/p>\n

Verizon’s value framework for GenAI use cases is built on three questions: Is the solution being used? Does it do what we want? Does it create value? The response metrics cover adoption monitoring, performance monitoring, and leading indicators of business outcomes.<\/p>\n

Ecosystem partnership instead of build or buy<\/h2>\n

Deutsche Telekom recognized that fine-tuning generic AI models for industry-specific applications is expensive and often produces suboptimal results when done in isolation. It developed a decision framework that evaluates when and why partnership is a better alternative to buying or building internally, then built an AI alliance with telecommunications industry peers. The goal is to offer a more uniform experience to global telecom customers, share LLM vendor costs, and collectively access capabilities that would be prohibitive individually. Each participant’s delivery teams collaborate to fine-tune models for the sector’s specific needs.<\/p>\n

Design patterns and TRiSM: governance embedded in code<\/h2>\n

Isbank, Turkey’s largest private bank, structured a “foundational AI product line” with AI squads aligned to critical capabilities such as retail pricing and marketing. Each squad includes data translators who act as customer success managers for central AI platforms, intercepting emerging needs from business teams and helping them maximize the value of available capabilities.<\/p>\n

Verizon manages governance not only through policies but by embedding appropriate controls in pre-approved workflows and templates: design patterns. Each pattern is calibrated for a family of use cases that perform similar tasks. The customer\/employee engagement pattern, for example, provides common guidance for all use cases that answer questions via Q&A \u2014 customer support, onboarding assistant \u2014 including input\/output specifications, model recommendations, prompting best practices, quality assurance guidelines, integration requirements, and compliance standards.<\/p>\n

The next level is the mechanization of AI policies through Trust, Risk and Security Management (TRiSM) technologies: “guardian agents” or AI policy engines \u2014 such as Bosch’s AI shield \u2014 that automatically enforce complex policies across all use cases without requiring manual implementation for each one.<\/p>\n

The message for CIOs is clear: AI does not scale by optimizing individual use cases. It scales by changing how the organization decides, learns, measures, and builds.<\/p>\n<\/section>\n","protected":false},"excerpt":{"rendered":"

Scaling AI in the enterprise does not just require new tools: it requires structural changes to the IT operating model. Five concrete practices adopted by CIOs who are already delivering results.<\/p>\n","protected":false},"featured_media":34448,"menu_order":0,"template":"","insights_category":[577],"insights_tags":[601,645,659,755,853],"class_list":["post-34446","insights","type-insights","status-publish","has-post-thumbnail","hentry","insights_category-strategy-and-ai","insights_tags-ai-governance-en","insights_tags-center-of-excellence-en","insights_tags-cio-en","insights_tags-it-operating-model-en","insights_tags-strategy"],"acf":[],"_links":{"self":[{"href":"https:\/\/askme.it\/en\/wp-json\/wp\/v2\/insights\/34446","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/askme.it\/en\/wp-json\/wp\/v2\/insights"}],"about":[{"href":"https:\/\/askme.it\/en\/wp-json\/wp\/v2\/types\/insights"}],"version-history":[{"count":1,"href":"https:\/\/askme.it\/en\/wp-json\/wp\/v2\/insights\/34446\/revisions"}],"predecessor-version":[{"id":34447,"href":"https:\/\/askme.it\/en\/wp-json\/wp\/v2\/insights\/34446\/revisions\/34447"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/askme.it\/en\/wp-json\/wp\/v2\/media\/34448"}],"wp:attachment":[{"href":"https:\/\/askme.it\/en\/wp-json\/wp\/v2\/media?parent=34446"}],"wp:term":[{"taxonomy":"insights_category","embeddable":true,"href":"https:\/\/askme.it\/en\/wp-json\/wp\/v2\/insights_category?post=34446"},{"taxonomy":"insights_tags","embeddable":true,"href":"https:\/\/askme.it\/en\/wp-json\/wp\/v2\/insights_tags?post=34446"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}