GenAI for EA: Welcome to a new world (2025)

0 Copyright © 2024 by Boston Consulting Group. All rights
reserved. William El Kaim, PhD IT Architecture Director BCG Platinion, Paris Blog, Speakerdeck, Linkedin Presenter today Source: Dall-E image generation GenAI for Architects: Welcome to a new world

Copyright © 2024 by Boston Consulting Group. All rights reserved.
1 Enterprises are facing many challenges, leading to seismic shifts in strategy Enabling customers and partners to compose and orchestrate end-to-end process flows from building blocks and execute them directly (and not only model or document them) on auto- pilot (with the help of technologies such as machine learning for example) Agentic AI is seen as an effective solution to create over the top (on top of existing ERP and business app) process orchestration with reasoning and self-learning capabilities Need to switch towards an ecosystem centered composable business approach Enterprises will build resilience by creating environments that enable businesses to build and consume business processes in a composable fashion and orchestrate them across system boundaries.

reserved. Build Composable architecture Build tech stacks by combining an assortment of smaller, reusable components serving different purposes in one unified system. This allows for great flexibility, scalability, and security. Adopt Generative and Agentic AI (Gen)AI and Agentic AI applications requires adoption of new patterns, technologies, and skills, further complicating the technology and security landscape for enterprise architects Revise your SDLC The advent of AI and more accessible development tools, commoditize soft. dev. and reshape the SDLC. It favors the shift towards "internal development centers", primarily to regain control, strengthen internal capabilities, improve service quality, and reduce cost Leverage cloud with frugality Adopting cloud-native architecture enhances agility, scalability, and reliability, enabling businesses to innovate faster. Objective is also to be frugal and deliver cost-aware, sustainable, and maintainable solutions (see: the laws of frugal architectures). 1 2 3 4 5 Source: BCG Platinion Composable business requires to build AI apps while optimizing legacy Source: BCG Platinion Be Data Driven Companies are becoming data-driven, but struggle with outdated systems, lack of efficient governance and resistant cultures. To improve maturity, consider data as a product deployed on composable data platforms

reserved. Composable architecture might become the keystone of modern software design Source: Gartner - https://www.gartner.com/en/supply-chain/insights/power-of-the-profession-blog/are-composable-manufacturing-systems-in-your-future Composable business architecture is a framework to maximize the ability to build, assemble and reassemble different business elements for the digital era. Business elements that can be composed include products, services, responses, experiences and organizations. The scope of the framework spans customer engagement, ecosystem partnerships and all operations. Composable architecture according to Gartner… SVC SVC SVC Package Business Capability (PBC) Optional User Interface API Event Channels Packaged business capabilities (PBCs) are application building blocks that have been purchased or developed internally & can deliver a specific business function, including microservices, APIs, or events Technology Build Composable architecture

reserved. Zoom| SAP relies on 5 architecture pillars for its composable enterprise Source: https://community.sap.com/t5/sap-teched-blog-posts/the-future-is-now-check-out-quot-architecting-the-future-sap-s-composable/ba-p/219972 Composable architecture and suites qualities are required to build composable Business Build Composable arch. S4Hana clean core and in-app and side-by-side extensions Revise your SDLC SAP Build, ABAP AI, Cloud ALM, SAP BTP Be Data Driven SAP Business Data Cloud Adopt Gen. and Agentic AI SAP Business AI, Joule Agent, Joule for Developers, etc. Leverage cloud with frugality SAP Grow and Rise 1 2 3 4 5 1 2 3 4 Technology Build Composable architecture

reserved. Shift from Application Driven to Data Driven … Build Data Products Data in platform and apps consume them Data Mesh, Stream processing, Data pipelines Build software applications Data created in apps and exchanged between apps Core systems, servitization (API), Event-driven, EAI • Apps created to deliver a process through a seamless user journey • Fast-paced iterations (days to weeks) to add & improve features • Evolves quickly over time to adapt to changing Business priorities From an "application-driven" approach… …Towards a "data-driven" approach Key area of focus • Focus on generating right quality data for applications and algorithms to refine into valuable Business insights • Automated treatment pipelines to refresh insights from minutes to real time • Stable in time (automatically adapting when Business data content evolves) Source: BCG Platinion Technology Be Data Driven

reserved. Adopt a product approach with industrialized delivery on composable data platform Focus Key practices Shift from project-based to product- oriented delivery Build modular and scalable data and (Gen)AI infrastructure Build scalable data product to become a data driven company • Define clear product ownership with cross-functional teams • Prioritize customer-centric outcomes over technical deliverables • Embed iterative feedback loops to continuously improve • Implement APIs and microservices for seamless data integration • Adopt a "plug-and-play" architecture to support diverse use cases • Enable self-service capabilities for business users to access and analyze data • Establish a framework to design, build, and deploy data products efficiently • Implement a governance model to manage the data product portfolios • Promote decision-making based at all organizational levels Product Approach Composable Data Platform/Fabric Industrialized Delivery Source: BCG Platinion Technology Be Data Driven

7 Zoom| SAP Business Data Cloud is data product native Technology Be Data Driven Source: SAP Business Data Sets Master and transactional objects/entities, analytical data sets, config data, etc. all semantically aligned Easily Consumable Simple SaaS access from data modeling and BI tools, while fully integrated with data engineering tools Well Described via rich metadata descriptions and including semantics Discoverable via an SAP data product catalog Data product properties SAP managed S/4HANA PCE Success Factors Ariba Concur … Custom managed Third-party S/4HANA ECC BW[4] Data integration & transformation Zero copy delta share SAP Business Data Cloud

reserved. Zoom| New North Star for SAP (feb. 2025 announcements) SAP LoB Source Systems SAP Managed S/4HANA PCE ISBN HCM CX Sustainability SAP Datasphere SAP BDC Cockpit Entitlement Management Data Product Installation Security and Governance SAP BW PCE SAP Analytics Cloud BI, Joule, Planning Insight Apps Object Store Standardizing data ingests for SAP LoBs SAP Data Products (based on canonical data model) Custom Data Products (based on S/4HANA Z-CDS-Views, Customer managed Source Systems, SAP BW, Partners or Databricks) Machine Learning Joule Knowledge Graph Data product generator Partner Connect Open data ecosystem Non-SAP Source systems Files API AWS, Azure, etc. SAP LoB Source Systems Customer Managed S/4HANA On-Prem BW ECC SAP Databricks* Custom AI / ML Bi-directional sharing Zero Copy Semantics Source: SAP Technology Be Data Driven

reserved. Right brain + left brain| Generative AI will co-exist with Predictive AI Not exhaustive Demand forecasting Protein design & selection Unstructured data ingesting & interpretation Creative content generation Synthesize findings in large datasets Other Generative AI applications Write and debug code Use Generative AI for content generation Generative AI Dynamic pricing engines Staff churn prevention Ad spend optimization Other Predictive AI/ML applications Fraud detection Use Predictive AI for decision-making Predictive AI/ML Use the combination of Predictive AI & GenAI to maximize impact generation Source: BCG Technology Adopt Generative and Agentic AI

reserved. Four key (Gen)AI breakthroughs in the past six months GPT-4o can reason across voice, vision & text in real time in the same model to interact seamlessly “voice to voice” Represents step towards a more natural human- computer interaction & AGI more broadly Likely to see desktop app start automating generic work tasks Easier and faster to build custom models for specific uses cases or audiences, significantly increasing accuracy for specific uses (vs. one-size-fits-all model) Could imagine specific models soon for BCG, for PAs, or potentially individuals More advanced models (Gemini Pro, GPT-4o) can retain context over a longer series of interactions, enhancing understanding & relevance of results More “memory” equals better performance and reduces needs for awkward workarounds Likely to see continued acceleration of cost reduction, speed increases GPT-4o is 2x faster & 50% cheaper vs GPT-4 Costs lowering faster than Moore's Law MSFT, others will likely build into platforms Edge-computing creating new hyper-responsive experiences (e.g. 4o’s upcoming responsive voice assistance, MSFT CoPilot + PC) Introduction of multi-modality Personalization through fine-tuning Increased memory and “context windows” Faster and cheaper 1. GPT-4o accepts ~16x more tokens than GPT-4 and Gemini Pro leads with 1M token "context windows” i.e., ~8,000 to 128,000 tokens Sources: ResearchGate, OpenAI, BCG analysis Technology Adopt Generative and Agentic AI

11 2025's biggest surprise so far: Reasoning is less of a moat than anyone thought Source: latent Space, https://www.latent.space/p/reasoning-price-war Chain-of-Thought (CoT) reasoning • Allows the model to break down complex problems into smaller, manageable steps. Reinforcement learning (RL) • Allows the model to learn from its own experiences, improving its decision- making over time. Technology Adopt Generative and Agentic AI

reserved. Perplexity| What can you tell me about my app landscape Technology Adopt Generative and Agentic AI Source: Perplexity.ai

reserved. Source: BCG Platinion Zoom| Archie, can you show me the ESG related data products? Fully autonomous agent, able to reason and to query and retrieve information from multiple internal and external sources Technology Adopt Generative and Agentic AI When combined with a ReACT agent, LeanIX evolves into a dynamic knowledge hub

reserved. Architects must train and leverage the "brain" for their tasks Business Architecture Data Architecture Application Architecture Technology Architecture Source: Jesper Lowgren, https://www.linkedin.com/pulse/enterprise-architecture-40-redefining-ea-age-ai-jesper-lowgren-qqiuc/ Technology Adopt Generative and Agentic AI

reserved. (Gen)AI and agentic AI app are a new generation of application architects will have to deal with A pervasive front- end layer 1 A data/semantic layer 3 A reasoning layer 2 • Frontends are being simplified, and AI Is becoming the new UI (see SAP — AI in 2025: Five Defining Themes). • User can access AI capabilities from their collaboration tool (like slack or Teams), from existing business applications (as a chat window or a copilot), or from a dedicated user interface. • Frontends support natively multimodal interactions (voice, chat, etc.) improving drastically the user experience • Set of (Gen)AI models that can be “static” and are directly usable (Large Language Model, Small Language Model) or “dynamic”, meaning they can evolve, learn and improve (for example ML models) • Both require to implement an additional layer, acting as an AI centric firewall, to ensure model will not drift or be misused (e.g., NVIDIA NeMo Guardrails) • Best results are obtained by associating predictive and generative AI models and orchestrating them as a predefined workflow (created for example with Langchain) or as a self-orchestrating component • Highly commoditized layers with open standards (Apache Iceberg, Databrick Delta Sharing for secure data sharing, etc.) • A “knowledge” or “semantic” layer can also be added to improve the accuracy and breadth of the AI solution (see Promotheux) • Data is more and more offered as a product and considered as an asset of a company (it has a fiduciary value) • Finally, multimodal data (structured and unstructured data) can now be leveraged fully and easily by the reasoning layer Technology Adopt Generative and Agentic AI

reserved. Gen(AI) Layer Knowledge Layer Copyright © 2022 by Boston Consulting Group. All rights reserved. 'Knowledge layer' capability, focused on business context, semantics, ontologies, is the next edge to ensure sharing of business context across applications Current implementation of LLM Powered applications Gen(AI) Layer Data product Layer Infrastructure Layer Data Layer Engagement layer Today Decoupled Engagement and Knowledge Layer Future Core transaction Layer .. .. .. Use Case specific Eng. layer L L M Core transaction Layer Infrastructure Layer Data product Layer ... … … Security and Access Security and Access Technology Adopt Generative and Agentic AI

18 (Gen)AI and agentic AI app are more difficult to design and run Data Models Code Guardrail • (Gen)AI and agentic AI app must cope dynamically with three assets: data, AI (foundation) models and code • Legacy application are more “static", since evolving data or software requires to build a new release New architecture patterns for Agentic AI MHQA = multihop question answering: ReAct = reason + act, RAG = retrieval- augmented generation, LLM = large language model. Source: Gartner (823011 C) New component architecture for Agentic AI LLM AI Agent Tools/plugins Backend Agent framework (e.g. LangChain, SemanticKernel) Tools/plugins Tools/plugins Tools/plugins Tools/plugins LLM for Actions External API/ systems Internal API Internal Systems Front End Memory Store/retrieve Execute Actions Technology Adopt Generative and Agentic AI

reserved. SAP Business AI offers provide a no/low-code tool for building agent Source: Based on the SAP News Center article "AI in 2025: Five Defining Themes," here is a synthesized slide outlining the key themes Technology Adopt Generative and Agentic AI

20 The rise of the builders … Source: Gartner, Andreessen Horowitz 1. Automated code generation: Tools like GitHub Copilot or Cursor enable developers to generate code snippets, functions, or even entire modules from high-level descriptions, reducing manual effort and errors 2. Enhanced UI/UX Design: Generative AI accelerates prototyping by creating wireframes, mockups, and even interactive designs quickly 3. Streamlined Testing and debugging: AI automates test case generation, identifies bugs, and predicts potential issues, ensuring higher code quality and faster iterations 4. Personalized user experience: By analyzing user behavior, generative AI tailor's app interfaces and content to individual preferences, driving engagement and satisfaction 5. Accelerated time to Market: Generative AI reduces development cycles, enabling faster prototyping, iteration, and product launches to meet market demands • Skills Gap: Only 24% of developers consider themselves proficient in generative AI • Code and deployment complexity: generated code are not easy to understand, modify and deploy • Technical Debt: Rapid code generation can lead to shortcuts that increase long-term maintenance costs • Security Risks: AI-generated code may introduce vulnerabilities like weak encryption or improper access controls • Ethical Concerns: Issues like bias in training data or copyright infringement require careful oversight • With the advent of artificial intelligence and more accessible development tools, creating sophisticated software solutions has become easier than ever before • This democratization of software development has led to a proliferation of options for businesses, driving down prices and reducing the perceived value of individual SaaS offerings • A new set of “over the top” AI platforms and solutions vendors are emerging and willing to disrupt, at term, the legacy SaaS application vendors • Organizations are bringing back software engineering in-house to "strengthen internal capabilities, improve service quality, regain control, and eliminate vendor markups Generative AI is reshaping the SDLC With still outstanding challenges to address … That do not code Technology Revise your SDLC By 2026, generative AI will significantly alter 70% of the design and development effort for new web applications and mobile apps

reserved. Zoom| With the advent of AI and more accessible development tools, creating sophisticated software solutions has become easier than ever before. Source: Abbas Khizer post, Henri Shi, El Kaim William StackBlitz Lovable Cursor OpenHands tempo Gemini DeepSeek AnthropicOpenAI OnLook Aider Vercel V0 GitHub Copilot Windsurf local AI-powered coding Non-technical user Native LLM Technology Revise your SDLC

reserved. Outcomes @ scale Right brain + left brain 10/20/70 Critical success factors to realize value • GenAI productivity gains are material – but require E2E transformation of operating models • A portfolio mindset is required as GenAI gain goes beyond productivity and can support new businesses • Going to PoC / demo is easy, scaling up is extremely hard – multiple challenges across tech, people, risks, economics, different by value play • Predictive AI offers decision & precision, and Generative AI leans into creation & dialogue • Engineering excellence in Generative AI and Predictive AI is key to success • Building AI systems is where true value comes from, incl Predictive AI, Generative AI, data fundamentals and integrating into core • 50%+ of workforce will be impacted in any given company • Given variety of off-the-shelf solutions, people are the critical path • It is about reskilling more than productivity • Wide difference of impact and significant risks to thought diversity • Need to build a large experiment & scale muscle in volatile environment Technology Success Factors Source: BCG

reserved. Three plays to win to get the full value out of (Gen) AI D E P L O Y GenAI in everyday tasks Accomplish 10-20% productivity increase by leveraging off-the-shelf GenAI tools and at-scale user training. Examples • Meeting summary • Code development • Calendar Management • Invoice reconciliation R E S H A P E Critical functions Reach 50%+ productivity boost with fit-for- purpose tech/data platforms and extensive workforce planning Examples • Marketing • Customer Service • Development / Engineering • Technology I N V E N T New business models Generate new revenue streams with bespoke systems and multi-functional incubation teams Examples • Reinventing customer experience • Insights & innovation platform • GenAI powered offerings Source: BCG, see Randy Bean 2025 Ai & Data Leadership Executive benchmark survey Companies are now focusing on aggressive use cases (vs. defensive) Technology Success Factors Outcomes @ scale

reserved. ROI Timeline for Generative AI use cases Source: The ROE of GenAI, Google Cloud, January 2025 Revenue Growth resulting from Generative AI use cases ROI Timeline for Generative AI use case Technology Success Factors Outcomes @ scale

reserved. of their AI investment to algorithms of their AI investment to embedding AI into business processes and new ways of working of their AI investment to technologies 10-20-70 rule: AI success requires a major business transformation effort Pioneers of AI at scale typically dedicate … Source: BCG & MIT analysis, report "How to Win with Artificial Intelligence" 10% 20% 70% Technology Generative AI 10/20/70

26 The services and materials provided by Boston Consulting Group
(BCG) are subject to BCG's Standard Terms (a copy of which is available upon request) or such other agreement as may have been previously executed by BCG. BCG does not provide legal, accounting, or tax advice. The Client is responsible for obtaining independent advice concerning these matters. This advice may affect the guidance given by BCG. Further, BCG has made no undertaking to update these materials after the date hereof, notwithstanding that such information may become outdated or inaccurate. The materials contained in this presentation are designed for the sole use by the board of directors or senior management of the Client and solely for the limited purposes described in the presentation. The materials shall not be copied or given to any person or entity other than the Client (“Third Party”) without the prior written consent of BCG. These materials serve only as the focus for discussion; they are incomplete without the accompanying oral commentary and may not be relied on as a stand-alone document. Further, Third Parties may not, and it is unreasonable for any Third Party to, rely on these materials for any purpose whatsoever. To the fullest extent permitted by law (and except to the extent otherwise agreed in a signed writing by BCG), BCG shall have no liability whatsoever to any Third Party, and any Third Party hereby waives any rights and claims it may have at any time against BCG with regard to the services, this presentation, or other materials, including the accuracy or completeness thereof. Receipt and review of this document shall be deemed agreement with and consideration for the foregoing. BCG does not provide fairness opinions or valuations of market transactions, and these materials should not be relied on or construed as such. Further, the financial evaluations, projected market and financial information, and conclusions contained in these materials are based upon standard valuation methodologies, are not definitive forecasts, and are not guaranteed by BCG. BCG has used public and/or confidential data and assumptions provided to BCG by the Client. BCG has not independently verified the data and assumptions used in these analyses. Changes in the underlying data or operating assumptions will clearly impact the analyses and conclusions. Copyright © 2024 by Boston Consulting Group. All rights reserved.

bcg.com

GenAI for EA: Welcome to a new world (2025)

GenAI for EA: Welcome to a new world (2025)

William El Kaim PRO

More Decks by William El Kaim

Other Decks in Technology

Featured

Transcript

0 Copyright © 2024 by Boston Consulting Group. All rights

Copyright © 2024 by Boston Consulting Group. All rights reserved.

2 Copyright © 2024 by Boston Consulting Group. All rights

3 Copyright © 2024 by Boston Consulting Group. All rights

4 Copyright © 2024 by Boston Consulting Group. All rights

5 Copyright © 2024 by Boston Consulting Group. All rights

6 Copyright © 2024 by Boston Consulting Group. All rights

Copyright © 2024 by Boston Consulting Group. All rights reserved.

8 Copyright © 2024 by Boston Consulting Group. All rights

9 Copyright © 2024 by Boston Consulting Group. All rights

10 Copyright © 2024 by Boston Consulting Group. All rights

Copyright © 2024 by Boston Consulting Group. All rights reserved.

12 Copyright © 2024 by Boston Consulting Group. All rights

13 Copyright © 2024 by Boston Consulting Group. All rights

14 Copyright © 2024 by Boston Consulting Group. All rights

15 Copyright © 2024 by Boston Consulting Group. All rights

16 Copyright © 2024 by Boston Consulting Group. All rights

17 Copyright © 2024 by Boston Consulting Group. All rights

Copyright © 2024 by Boston Consulting Group. All rights reserved.

19 Copyright © 2024 by Boston Consulting Group. All rights

Copyright © 2024 by Boston Consulting Group. All rights reserved.

21 Copyright © 2024 by Boston Consulting Group. All rights

22 Copyright © 2024 by Boston Consulting Group. All rights

23 Copyright © 2024 by Boston Consulting Group. All rights

24 Copyright © 2024 by Boston Consulting Group. All rights

25 Copyright © 2024 by Boston Consulting Group. All rights

26 The services and materials provided by Boston Consulting Group

bcg.com