Generative AI Enterprise
Posts
KPMG: Value at stake in 17M+ companies

KPMG: Value at stake in 17M+ companies

Plus, Microsoft agents, McKinsey on sales growth, and more.

Lewis Walker
March 30, 2025

WELCOME, EXECUTIVES AND PROFESSIONALS.

Many enterprises face a pivotal shift from piloting to scaling gen AI. With mounting pressure to quantify returns and justify investment, KPMG analyzes 17M+ companies worldwide to uncover the value at stake.

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:

KPMG quantifies the gen AI opportunity.
McKinsey: Unlocking profitable B2B sales growth.
Microsoft introduces Researcher and Analyst agents.
Deloitte: Operationalizing AI governance.
Google launches Gemini 2.5 Pro.
Transformation and technology in the news.
Career opportunities & events.

Read time: 4 minutes.

VALUE ASSESSMENT

KPMG quantifies the gen AI opportunity

Image source: KPMG

Brief: To help clients accelerate AI transformations, KPMG spent 18 months analyzing 3 billion+ data points for 17 million+ companies worldwide to quantify the value at stake from fully deploying and adopting gen AI.

Breakdown:

The gen AI opportunity equates to 4-18% of EBITDA or 19-23% of salary cost annually in labor productivity alone across sectors.
Productivity gains were quantified by estimating freed-up time, multiplied by salaries, assuming reinvested time is equally productive.
48% stems from low/medium complexity tasks, while 52% involves high-complexity work requiring tailored solutions, agents and complex change.
EBITDA impact varies by sector, with the highest in Professional Services (17.6%), Technology, Media & Telecom (15.5%), and Healthcare (14.6%).
Front Office (excl. Sales & Marketing) holds 22.4% of gen AI’s value. IT (16%), Sales (15.8%), Supply Chain (15.4%), Finance (9.3%), and Marketing (6.4%).

Why it’s important: Executives remain willing to invest in gen AI for productivity and competitive advantage, but often face challenges in quantifying expected returns and setting realistic targets. KPMG’s value assessment helps address these challenges, shaping strategy, guiding investments and accelerating transformation.

BEST PRACTICE INSIGHT & CASE STUDIES

McKinsey: Unlocking profitable B2B sales growth

Image source: McKinsey & Company

Brief: McKinsey explores how gen AI drives profitable B2B sales growth, analyzing seven key use cases across the deal cycle, their impacts on sales ROI and customer experience, and real-world deployments by leading organizations.

Breakdown:

19 percent of enterprises have implemented gen AI use cases for B2B buying and selling, and another 23 percent are in the process of doing so.
Opportunities include identifying and prioritizing leads, enabling outreach, optimizing pricing, supporting research, and coaching sellers at scale.
Without AI, for instance, sales teams rely on fragmented lists with no clear guidance to prioritize opportunities or help predict customer actions.
With AI, a prioritized list of customer-specific recommendations can help sellers target opportunities, improving efficiency and conversion rates.
A firm highlighted leveraged gen AI to personalize outreach, adding $1B in opportunities, growing its pipeline 10%, and doubling click-through rates.

Why it’s important: Most enterprises are in the early stages of empowering teams to harness gen AI and increasingly autonomous agents for better insights and higher conversion rates. With the right strategy, go-to-market model, and decisive action, B2B leaders can unlock enormous potential.

INNOVATION INSIGHT

Microsoft introduces Researcher and Analyst agents

Image source: Microsoft

Brief: Microsoft launched two reasoning agents: Researcher and Analyst. They securely access emails, meetings, files, and chats, along with third-party integrations and the web, to analyze information and deliver insights on demand.

Breakdown:

Researcher combines OpenAI’s deep research with Microsoft 365 Copilot’s orchestration, search and data integration capabilities.
For instance, Researcher can quickly build a detailed go-to-market strategy using proprietary data and emerging external trends.
Analyst is built on OpenAI’s o3-mini reasoning model leveraging chain-of-thought reasoning for advanced data analysis.
For example, Analyst can turn raw data scattered across multiple spreadsheets into a visualization of customer purchasing patterns.
Researcher and Analyst will roll out to enterprise customers in April as part of Microsoft 365 Copilot's early-access Frontier Program.

Why it’s important: When the latest technology meets Microsoft’s enterprise reach, impact is almost inevitable. Nearly 70% of Fortune 500 companies use Microsoft 365 Copilot. According to CEO Satya Nadella, these agents “will completely change the scale and scope of what any one of us can do at work."

BEST PRACTICE INSIGHT

Deloitte: Operationalizing AI governance

Image source: Deloitte

Brief: Deloitte published a 12-page report on operationalizing AI governance, emphasizing that success lies in maintaining control without sacrificing innovation, ensuring rigorous yet efficient governance.

Breakdown:

Any technology carries risks, and AI is no exception, such as the potential for unreliable yet convincing outputs and autonomous decision-making.
Regulations like the EU Act must be operationalized without burdensome red-tape that would strip away many of the benefits of utilizing AI.
Two key AI governance components are QMS, ensuring AI builds meet quality standards, and RMS, which monitors AI to resolve risks early.
Yet, AI governance extends beyond quality and risk management, spanning structures, practices, processes, and systems to ensure compliance.
RMS is a prime example of "systems" in action, linking risk monitoring directly to AI models to ensure governance remains nimble and responsive.

Why it’s important: Enterprises increasingly rely on AI to maintain their competitiveness, and rapid proliferation necessitates professional management of quality, risk, and compliance. Effective governance should not just enforce controls but also be focused, lean, efficient and effective.

INNOVATION INSIGHT

Google launches Gemini 2.5 Pro

Image source: Google

Brief: Google launched Gemini 2.5, a new family of AI reasoning models. The first release, 2.5 Pro, is an experimental version that tops multiple key benchmarks and debuts at #1 on the LMArena Leaderboard.

Breakdown:

On Humanity's Last Exam benchmark (reasoning & knowledge), Gemini 2.5 Pro leads with 18.8%, surpassing OpenAI o3-mini's 14%.
It also leads on GPQA Diamond (science), ANIME 2025 (math), and MMMU (visual reasoning) with scores of 84%, 86.7%, and 81.7%, respectively.
On SWE-Bench Verified, the industry standard for agentic coding, Gemini 2.5 Pro performs strongly at 63.8%, trailing Claude 3.7 Sonnet's 70.3%.
The model ships with a 1M token context window, set to double to 2M for processing large datasets and entire code repositories.
Gemini 2.5 Pro is available now in Google AI Studio and the Gemini app for Gemini Advanced users, and will be coming to Vertex AI soon.

Why it’s important: Google continues to push state-of-the-art models, but with GPT-5 on the horizon, it’s uncertain how long Gemini 2.5 Pro will maintain its top spot. Nonetheless, it's a significant upgrade for enterprises, as Google commands 39% of the market for "foundation models and model management platforms."

Capgemini’s 112-slide report explores AI’s impact on customer service, with 70% of agents seeing reduced workloads from generative and agentic AI.

BCG stresses the need for proper human oversight in gen AI, stating that "vibes over facts" is not a viable approach to mitigating AI-related risks.

Bain examines AI's impact on reshaping company competition, outlining nine key themes that define the evolving enterprise AI landscape.

PwC launched Agent OS, an AI command center that connects and scales intelligent agents into workflows "up to 10x faster than traditional methods."

Deloitte explores AI's impact on work and how organizations can create an employee value proposition (EVP) that makes AI a friend, not a foe.

Citi Group hosted Greg Ulrich, Mastercard’s Chief AI and Data Officer, for a discussion on the potential and challenges of agentic AI.

Booz Allen provides insights on AI for IT Operations (AIOps), tackling key challenges human operators face when working alongside AI agents.

Andreessen Horowitz explores ‘God Mode’ shopping, where AI will predictively bring products to you at the right price and size.

OpenAI will adopt Anthropic’s Model Context Protocol, extending integration options with external data and software for ChatGPT and other products.

Anthropic’s new research papers reveal how Claude processes data, improving understanding of its reasoning, multilingual skills, and advanced planning.

OpenAI added image generation to GPT-4o and Sora, shifting from separate text and image systems to a unified approach for more precise, context-aware visuals.

xAI acquired X (formerly Twitter), merging data distribution and talent in a deal valuing xAI at $80B and X at $33B ($45B minus $12B debt).

DeepSeek released an updated version of its V3 model, featuring a highly permissive MIT open-source license for broad use and customization.

Kai-Fu Lee said, “Sam Altman is probably not sleeping well,” as his startup 01.AI pivots to DeepSeek’s open models at just 2% of OpenAI’s annual costs.

Butterfly Effect, the startup behind Manus AI, is seeking funding at a $500M valuation as it confronts massive cash burn from Claude API costs.

UiPath’s Agentic AI Summit recap highlights its latest agentic development, orchestration testing capabilities, and how enterprises are leveraging AI agents.