Generative AI Enterprise
Posts
IBM: The AI prize that truly matters

IBM: The AI prize that truly matters

Plus, McKinsey workforce evolution, OpenAI case study, and more.

Lewis Walker
March 02, 2025

WELCOME, EXECUTIVES AND PROFESSIONALS.

Productivity is important, but transformative innovation is the ultimate prize. IBM explores how leading enterprises are harnessing AI as a catalyst for innovation, competitiveness, and growth.

Since the previous edition, we've reviewed hundreds of the latest insights on best practices, case studies, and innovation. Here’s the top 1%...

In today’s edition:

IBM: What sets AI transformation leaders apart.
McKinsey: Workforce planning in the age of gen AI.
Financial analysis with OpenAI's reasoning models.
OpenAI releases GPT-4.5.
Transformation and technology in the news.
Career opportunities & events.

Read time: 4 minutes.

EXECUTIVE INSIGHT & CASE STUDIES

IBM: What sets AI transformation leaders apart

Image source: IBM

Brief: IBM published a 48-page report on what AI transformation success looks like across 10 industries, featuring enterprise case studies, executive perspectives, and recommended actions for each industry.

Breakdown:

Enterprises fixated on efficiency risk missing AI’s true potential: innovation, growth, and breakthroughs once thought not possible.
Achieving leadership requires an operating model that is people-centered, process-driven, technology/data-enabled, and partner-energized.
The report details what AI transformation success looks like across Finance, Retail, Telecom, Automotive, Healthcare, and other industries.
It features case studies and perspectives from NatWest, Bharti Airtel, NASA, Honda, the UK NHS, Informed Genomics, and more.
For instance, NatWest’s Cora+ AI, leveraging RAG, addresses over 11 million customer inquiries annually, improving satisfaction by 150% through more nuanced, anticipatory responses.
Wendy Renshaw, Chief Digital Officer at NatWest, emphasizes the importance of customer care, whether through human or digital interactions.

Why it’s important: Most CEOs recognize that leaders in gen AI, who go beyond productivity to drive growth, will capture greater competitive advantage. Successfully leveraging AI will be a defining factor in shaping the rise, or fall, of businesses across virtually all industries.

BEST PRACTICE INSIGHT

McKinsey: Workforce planning in the age of gen AI

Image source: McKinsey & Company

Brief: McKinsey explores five best practices for building a holistic talent plan through strategic workforce planning (SWP) in the age of gen AI, as up to 30% of existing work hours may be replaced through automation by 2030.

Breakdown:

Prioritize talent as much as capital. Employees are both the largest investment and greatest asset. Benchmarking against top peers helps set strategy.
Prioritize capacity and capability by using scenario-based supply and demand forecasting to identify talent gaps for hiring, development, and growth.
Balance external hires with redeployment, upskilling, reskilling, acquisitions, and outsourcing, particularly important amid the rapid pace of change.
Strategic workforce planning should be a continuous process, not a one-time fix. Leading firms align SWP with finance, operations, and HR.
McKinsey highlights examples, including a North American software firm that modeled gen AI adoption scenarios to guide workforce evolution.

Why it’s important: S&P 500 companies that excel in maximizing return on talent generate 300% more revenue per employee than the median firm. SWP is not a novel concept, but the rapid pace of gen AI reshaping work dynamics makes getting it right even more important.

AI-NATIVE CASE STUDY

Financial analysis with OpenAI's reasoning models

Image source: OpenAI

Brief: OpenAI published a case study on its work with Endex, a company developing an AI financial analyst. By integrating OpenAI’s reasoning models, Endex is achieving enhanced performance in tasks requiring structured thinking and deep analysis.

Breakdown:

Endex previously used complex prompts, chained completions, and verification steps. With OpenAI o1, it's now simpler without sacrificing accuracy.
With OpenAI o3-mini, Endex gains 3x faster intelligence, enabling multi-step workflows like automating financial model reconciliation.
Endex identifies discrepancies in financial data, flagging restatements and inconsistencies with citations, freeing analysts time for decisions.
OpenAI’s o1 vision capabilities allow Endex to process investor presentations, internal decks, Excel models, and 8-Ks, enhancing analysis.
Endex automates detailed reports, reducing manual financial analysis, letting professionals focus on strategy instead of data formatting.
Blind user testing shows financial analysts prefer OpenAI o1’s responses 70% of the time over the non-reasoning models in the chart above.

Why it’s important: Finance professionals require structured, referenceable reasoning, a challenge for non-reasoning LLMs. OpenAI's o-series models with long-context windows, prompt adherence, and advanced reasoning capabilities consistently achieved this for Endex.

INNOVATION INSIGHT

OpenAI releases GPT-4.5

Image source: OpenAI

Brief: OpenAI launched GPT-4.5 (Orion), its largest model yet. By prioritizing unsupervised learning over reasoning, it's more nuanced in communication, and less prone to hallucinations across a wide range of tasks.

Breakdown:

GPT-4.5 offers more natural conversation, improving understanding of human intent and emotional intelligence for smoother interactions.
The model hallucinates less than previous versions (37.1% vs. GPT-4o's 61.8%), with testers favoring it over GPT-4o for professional and creative tasks.
It outperforms o3-mini and o1 on SWE-Lancer, OpenAI's new coding benchmark, but doesn't advance in math or science from reasoning models.
While OpenAI's o-series prevented GPT-4.5 from dominating benchmarks, it is smarter than its non-reasoning predecessor, GPT-4o.
It is available in Azure AI Foundry and GitHub Copilot Chat for Copilot Enterprise users, and is rolling out to ChatGPT Enterprise plan users.
GPT-4.5 is 15x more expensive than GPT-4o, with reductions in hallucinations costing significantly more than previous leaps in intelligence.

Why it’s important: This isn’t an intelligence upgrade, but a trust upgrade, and it doesn’t come cheap. For regular users, GPT-4.5’s pricing lacks competitiveness. But for OpenAI’s 2M enterprise customers, for instance, in certain legal, healthcare and finance precision use cases where hallucinations are costly, it can be worth it.

McKinsey published an article on scaling AI-native telcos, including leveraging agentic AI across the enterprise. The strategy consultants also considered how AI infrastructure could be a new growth avenue for telcos.

IBM explores AI agent expectations vs. reality across four narratives: 2025 as the year of agents, agents handling complex tasks, AI orchestrators governing networks, and agents augmenting human workers.

Accenture and UK Finance shared a 59-page report on gen AI in financial services, covering adoption, trends, use cases, regulations, and detailed case studies on customer complaints, KYC, and software development.

Deloitte, NVIDIA, and ServiceNow released a 64-page study on AI practices in 250 cities across 78 countries. The study examines how cities are using AI and automation, from machine learning and RPA to generative and agentic AI.

Salesforce explores "jagged intelligence," where AI excels at complex tasks but struggles at times with simpler ones, posing challenges for enterprises that require reliable solutions.

Microsoft launched Phi-4 multimodal and Phi-4 mini SLMs, outperforming models twice their size in select tasks. It also announced new customization tools, and enterprise agent upgrades in Azure AI Foundry.

Anthropic is set to raise a larger than planned $3.5B funding round at a $61.5B valuation, following the recent release of Sonnet 3.7, highlighting continued growth and investor interest.

DeepSeek claims $200 million/year in projected revenue at a 545% profit margin, driven by efficient use of NVIDIA H800 GPUs costing just $87,000 daily, while charging just $2.19 per million tokens.

IBM unveiled the Granite 3.2 model family, offering compact reasoning, vision-language, and time series models tailored for enterprise use.

Meta reportedly plans to release a standalone Meta AI app in Q2, with potential paid subscription options, in a move to compete with OpenAI’s ChatGPT app.