Generative AI Enterprise
Posts
Deloitte: What board directors need to know

Deloitte: What board directors need to know

Plus, Google on AI's future, agentic automation, and more.

Lewis Walker
February 26, 2025

WELCOME, EXECUTIVES AND PROFESSIONALS.

We are at a pivotal moment in the history of human invention. But how can we balance the desire for rapid innovation with the patience to scale AI in a responsible and trustworthy manner?

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:

Deloitte: What board directors need to know.
Google: Perspectives on the future of AI.
Anthropic launches hybrid reasoning model.
Powering automation with agents.
Transformation and technology in the news.
Career opportunities & events.

Read time: 4 minutes.

STRATEGIC AI GOVERNANCE

Deloitte: What board directors need to know

Image source: Deloitte

Brief: Deloitte published its AI Governance Roadmap, a 16-slide guide, to help boards understand their role in AI governance. It explores key questions and resources for overseeing AI, regardless of the organization's AI maturity.

Breakdown:

Evaluate AI approach within corporate strategy, oversee execution, and help management adapt strategy to AI risks and opportunities as needed.
Oversee AI risks (strategic, functional, and external) to the company’s overall strategy, integrating them into the enterprise risk program.
Define oversight ownership at the board level (e.g. the full board, existing committee, new committee, or sub-committee). Consider an AI advisory council or adding an AI-expert board member.
Monitor performance against AI-specific strategic, financial, and operational goals. Establish a consistent evaluation process.
Assess if management has the skills to execute the AI strategy. Understand AI’s impact on recruitment, development, and incentives.
Cultivate trustworthy AI with appropriate disclosures and communications. The board tracks AI usage, ensuring adherence to ethical standards.

Why it’s important: Although AI is not new, the increasingly proliferation of gen AI in enterprises brings governance topics to the forefront. The decisions leaders make today in balancing opportunity and risk will shape their enterprises and society at large for years to come.

AI-NATIVE INSIGHT

Google: Perspectives on the future of AI

Image source: Google

Brief: Google’s 75-page report explores how startups are pushing the boundaries of AI innovation. It features insights from 23 entrepreneurs and venture capitalists (VCs) on where gen AI might be headed, along with the enterprise landscape.

Breakdown:

Glean Founder & CEO Arvind Jain says AI is about "doing things that weren’t possible before," and not just efficiency but growing your topline.
To create lasting value, products must be stickier, "indispensable and deeply integrated into the user’s workflow," states VC Crystal Huang.
Harrison Chase, CEO of LangChain, envisions agentic systems as “ambient agents,” always on in the background, and alerting only when needed.
Success will come from "network effects that lock-in access to data or customers,” according to David Friedberg, CEO of Ohalo Genetics.
Douwe Kiela, CEO of Contextual AI, says: "You get a head start if you assume the technology is ready, even if it’s not quite there yet."
“Your competition is no longer against the incumbent, it is against the incumbent’s business model,” asserts VC Jerry Chen.

Why it’s important: By understanding today’s AI-native startups, enterprises can better anticipate and plan for tomorrow, whether new practices, partners, or competitors. Founders building and the investors backing AI’s future offer valuable insights into what’s on the horizon.

INNOVATION INSIGHT

Anthropic launches hybrid reasoning model

Image source: Anthropic

Brief: Anthropic released Claude 3.7 Sonnet, its most advanced model and the first “hybrid reasoning” AI, blending controllable instant responses with extended thinking. Available on all Claude plans, Anthropic API, and select cloud platforms.

Breakdown:

Claude 3.7 Sonnet lets users switch between standard and “extended thinking" modes, with the latter revealing the AI’s reasoning.
API users can control how long Claude thinks (up to 128K output tokens), balancing speed, cost, and quality based on task complexity.
The AI achieves SOTA performance on coding benchmarks and agentic tool use, outperforming competitors like o1, o3-mini, and DeepSeek R1.
Anthropic introduced Claude Code in research preview, a command-line coding agent that edits files, reads code, and writes/runs tests.
Claude 3.7 Sonnet is now available on all Claude plans (Free, Pro, Team, Enterprise), Anthropic API, Amazon Bedrock, and Google Cloud’s Vertex AI.
The 43-page system card for this release details measures implemented to reduce harms via model training and safeguards.

Why it’s important: Anthropic’s enterprise market share grew from 12% to 24% in 2024. Enterprises can now leverage enhanced performance, especially in coding, while the hybrid approach enables more seamless workflows, combining speed, efficiency, and reasoning to tackle complex, multi-step problems when necessary.

BEST PRACTICE INSIGHT

Powering automation with agents

Image source: PwC

Brief: PwC released a 20-page report on Agentic Powered Automation (APA), providing leaders and AI professionals insights into its benefits, enterprise use cases, executive perspectives, and guidance on aspects of governance.

Breakdown:

APA combines gen AI agents, the brain, understanding requests and planning actions, with automation (e.g. APIs, RPA), the hands, executing actions.
The agents validate outputs throughout, and learn from results to improve performance over time.
The report highlights 20 agent use cases across finance, supply chain, sales, marketing, IT, and HR. For example, expense reconciliation in finance.
A HR talent search deep dive compares the as-is process vs. agentic solution (image above), highlighting how roles shift from transactional to strategic.
Data & Analytics Leader Sudipta Ghosh mentions the need to balance agentic innovation and compliance, among other PwC perspectives.

Why it’s important: The market is filled with terms: agentic AI, agentic-powered automation, or autonomous agents. Automating knowledge work isn’t new; but it can be faster and more expansive with gen AI, increasing agency for fuller, end-to-end automation and augmenting distinct human capabilities.

Bain shared how AI skills demand grew 21% annually since 2019, salaries by 11%, but talent remains scarce. Executives cite this as hindering gen AI implementation and adoption, with the gap expected to persist through 2027.

Accenture published an article on gen AI in commercial payments, covering the evolution of the industry, a use case taxonomy mapping enterprise opportunities from low to high maturity, and where to begin.

PwC, with IDC, released a 13-page report on transforming procure-to-pay with PwC’s Agentic AI Intelligent Spend Management Suite. It covers pain points, client expectations, benefits, design principles, and more.

Deloitte published 16 slides on navigating the EU AI Act, including compliance strategies. It details examples of unacceptable AI practices, steps for identifying risks, and an approach to governance, awareness, and monitoring.

Cisco shared 59 slides on mitigating AI investment risks with Cisco AI Defense. It covers the AI risk landscape, security frameworks, key threats, and an introduction to Cisco’s solution.

OpenAI rolled out its Deep Research feature for ChatGPT Plus, Team, Edu, and Enterprise tiers. The 35-page system card details how Deep Research was built, including risks assessed, and safety improved.

Google and Salesforce expanded their partnership, integrating Gemini into Agentforce. Salesforce Service Cloud will integrate more closely with Google Customer Engagement Suite, enhancing AI-powered contact center capabilities.

IBM plans to acquire DataStax to enhance IBM's watsonx portfolio of products helping unlock value from the vast amounts of unstructured data in enterprises and accelerate gen AI adoption.

DeepSeek is reportedly expediting the launch of R2, the successor to its January’s R1 model. Initially planned for May, the release timeline has been accelerated with the view to capitalize on momentum.

NVIDIA shared a framework for agentic autonomy, defining three levels of autonomy. It explores how autonomous systems can be manipulated, identifies potential threats, and recommends security controls for each level.