Edition in partnership with

Welcome executives and professionals. For board directors, the challenge is no longer whether AI matters, but how to oversee its use confidently, responsibly and effectively.

Since the previous edition, we have reviewed hundreds of the latest insights in agentic and generative AI, spanning best practices, case studies, market dynamics, and innovations.

This briefing outlines what is driving material value — and why it’s important.

In today’s briefing:

• AI priorities in the boardroom.

• AI agent cyber risks and controls.

• The value of long-running agents.

• AI is everywhere except the P&L.

• Transformation and technology.

• Insights for Executive+ members.

• Career opportunities & events.

Read time: 4 minutes.

Note: Enterprise AI Executive is published on Wednesdays only. We no longer publish on Sundays.

MARKET & BEST PRACTICE INSIGHT

AI priorities in the boardroom

Brief: KPMG’s former Board Committee Chair and AI Sales Acceleration Leader, Per Edin, drew on his interactions with over 1,000 board directors in the last three years to outline five AI priorities dominating boardroom agendas.

Breakdown:

• Board questions fall into five areas, starting with how to make sense of AI market trends and flavors like agentic and physical AI.

• Directors ask how to quantify, prioritize and capture the types of value AI can bring, and how to manage risks like cybersecurity.

• They also ask how to evolve corporate strategy, shape a transformation program and change how the board itself works in the AI era.

• The insights highlight KPMG accelerators, including its Rapid Value Assessment and AI Governance Principles for Boards.

Why it’s important: As organizations move from exploring AI to transforming and operating with it, the board’s role has never been more crucial or complex. AI brings real value and risk, and while aligning on a strategy isn’t always easy, executing enterprise-wide change is harder still.

IN PARTNERSHIP WITH TELEPORT

Your AI has more access than your people

Brief: Teleport research found 70% of organizations grant AI more access than humans. Over-privileged AI carries a 76% incident rate, least-privileged access cuts that 4.5x. The risk isn't the AI; it's the access behind it.

With Teleport, you get:

• Least-privileged access enforced per request for every AI identity

• Short-lived certificates that eliminate static credentials and secrets

• Full identity attribution for every AI action across your infrastructure

• Audit-ready visibility into human and machine access alike

See where your enterprise stands. Read the research.

BEST PRACTICE INSIGHT

AI agent cyber risks and controls

Brief: PwC’s 48-page report examines how AI agents that plan, connect to business systems and then act on their own create new cyber risks, and sets out practical governance controls matched to each agent’s level of autonomy.

Breakdown:

• As agents go from assistive (L1) to fully autonomous (L5), the focus moves from checking outputs to governing the actions they take.

• In two tests, prompt injection hidden in support tickets and passport images led agents to misuse broad privileges and leak data.

• PwC’s AI Control Assurance Level (AI-CAL) sets minimum preventive, detective and responsive controls for each agent autonomy level.

• Three priorities stand out: least-privilege access for agents, logging with the ability to halt them, and AI-CAL checks before go-live.

Why it's important: These three priorities do not lose their effectiveness even as foundation models, frameworks or regulations change. What enterprises should do is not attempt to fully predict future uncertainty, but rather build operations capable of adapting to change.

BEST PRACTICE INSIGHT

The value of long-running agents

Brief: Andreessen Horowitz’s David Haber argues long-running agents can now carry customers through multi-week journeys like insurance claims, giving vendors more opportunity to share in client revenue, not just cost savings.

Breakdown:

• Many complex customer journeys work like a relay race: most value is realized at the finish and can leak at each handoff on the way.

• Unlike single-task agents, long-running agents keep a case’s full history, act ahead of deadlines and run parallel sub-agents.

• In plaintiff law, agents can respond to new callers at any hour, keep following up with witnesses and review case documents in full.

• Vendors can sell long-running agents as software, take over outsourced work or run the business themselves, such as collecting debt.

Why it’s important: Long-running agents can help expand AI’s addressable market from labor budgets to a share of revenue. Outcome-based pricing can lift contract values above cost-cutting deals, but procurement habits, attribution disputes and fee-sharing rules are hurdles.

AI-NATIVE PROFESSIONAL

How to review a financial model with ChatGPT

Brief: In this guide you’ll learn how to have ChatGPT clean up your financial model and return a severity-ranked QA memo on formula risks, broken links, figures that don’t match their sources, and assumptions that need an owner.

Step-by-step:

1. Attach the financial model and its supporting source files, and keep the original workbook available so you can compare every edit.

2. Identify the audience, priority tabs, expected checks, and outputs, plus any protected assumptions ChatGPT must leave unchanged.

3. Run the starter prompt and request a cleaned copy and QA memo flagging hardcodes, stale links, period mismatches, and failed checks.

4. Open the workbook in ChatGPT, inspect changed formulas, links, and output tabs, and resolve owner questions before accepting edits.

Best practice: Use the follow-up prompt to verify fixes; the memo should rank issues by severity, cite cells, and flag ambiguous business logic.

For the full guide, including prompts, upgrade to Executive+ or The Boardroom.

MARKET & BEST PRACTICE INSIGHT

AI is everywhere except the P&L

Brief: Oliver Wyman’s 91-page report, drawing on 25,000+ survey responses in 15 countries and 415 CEOs, finds AI is everywhere except the P&L, as the technology is arriving faster than the decisions needed to turn it into value.

Breakdown:

• Leaders must make choices many have deferred, starting with what return each AI investment should produce and who will capture it.

• Decide AI’s role before asking staff to transform around it: only 11% of non-managers say leaders have set out a clear AI ambition.

• Decide how to win when machines help customers choose, as 48% of Gen Z and 40% of millennials now turn to AI first when shopping.

• Decide how to govern AI at the speed it operates, and which AI dependencies are acceptable and which are worth paying to control.

Why it’s important: All five reality checks come down to one management challenge: harnessing a technology that creates options faster than organizations can decide what to do with them. AI will keep improving whether or not leaders decide correctly. The returns will not.

McKinsey detailed how an AI operating system can become a company’s “second brain,” and the economics of applying AI to existing work.

BCG surveyed 398 CROs and sales leaders on AI-driven sales growth, and published a 26-page report rating over 75 economies on AI readiness.

IBM published a 15-page report on AI agent governance, finding 80% of executives say an AI incident reached their board or C-suite.

Anthropic published a 20-page study on AI’s impact on its workforce and a 16-page report on public vs frontier lab AI expectations.

Bain explored the reliability paradox: AI cuts the cost of building software but can raise the cost of running it unless ops change.

EY published a 16-page report on using synthetic worlds to rehearse AI decisions, and a report on CFOs managing AI economics.

Deloitte published a 13-page report on managing AI as an asset over its lifecycle, and one on why AI success needs new partners.

OECD published a 36-page report on agentic AI in organisations, with examples of AI agents built and deployed by 25 organizations.

Meta released Muse Spark Open, the most capable US open-weight model yet, with 816B parameters built for coding and agentic work.

Anthropic announced it is giving six METR researchers employee-level access to assess its AI safety practices and publish findings.

Deloitte launched Ascend Agentic Harness, built on Claude, adding human and automated checks to AI-assisted software development.

OpenAI released GPT-Live-1.1 and upgraded its Chrome extension to automate tasks, while countering Apple’s trade secret lawsuit.

Anthropic announced a Public Benefit Fund to get 5% of its equity (~$48B), and Elon Musk pledged a SpaceX share to every US child.

Bank of America reportedly hired about 1,000 contractors to patch software vulnerabilities uncovered by Anthropic’s Mythos and other AI models

The Financial Times reported Anthropic’s Q3 revenue hit $16.8B, up 45% QoQ, as it eyes an IPO that could value the company at $2T.

Reuters explained why CoreWeave is central to AI bubble fears, as heavy debt and doubts over its $104B order book lift bond yields.

CAREER OPPORTUNITIES

Mars - Global Head of AI & Data

JPMorganChase - AI Transformation Executive Director

AWS - Principal Executive AI Strategist

EVENTS

C-Vision - Agentic AI Executive Dinner - November 19, 2026

Forvis Mazars - Board AI Lessons - November 24, 2026

Snowflake - Scaling Enterprise AI Agents - December 03, 2026

Reach enterprise AI decision-makers:

• 66% of readers are C-level executives or VP and Director-level leaders.

• 63.2% of the audience is based in the U.S., EU, UK, ANZ, and Singapore.

• Read by leaders at Microsoft, Deloitte, the Fortune 500, and more.

Guaranteed impression and custom sponsorship packages available, with post-send performance reporting.

Conceived as a practical communication for executives Lewis Walker has worked with, this briefing has become a trusted resource for thousands of senior decision-makers shaping the future of enterprise AI.

We welcome your feedback.