Edition in partnership with

Welcome executives and professionals. British workers are spending close to £1 billion a year from their own pockets on generative AI tools they use for work.

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:

• 25,000 workers on GenAI at work.

• Enterprise AI spend takes priority.

• The reprioritization of IT services.

• The harness is not enough.

• Transformation and technology.

• Insights for Executive+ members.

• Career opportunities & events.

Read time: 4 minutes.

MARKET INSIGHT

25,000 workers on GenAI at work

Brief: In the largest single-country study of workplace GenAI use, Deloitte asked 25,000 UK workers how they use the technology at work and what they think about its growing role in their day-to-day working lives.

Breakdown:

• 63% of UK workers have used GenAI, but half have received no training and 65% say leaders discuss it without a clear understanding.

• Shadow AI is widespread: almost one in three GenAI users (31%) use the technology at work without their employer knowing.

• One in six GenAI users (17%) pay for their own AI tools that they use at work, at a cost to UK workers of £958 million a year.

• Workers say they save 70 minutes a week on average, yet nearly a quarter (23%) still sense a stigma attached to using GenAI at work.

Why it’s important: GenAI use is widespread but shallow. Many have tried it, few use it daily, and most stick to familiar tasks such as writing emails and searching for information rather than rethinking how work is done. Its potential needs investment in people and process, not just technology.

IN PARTNERSHIP WITH THE BOARDROOM

AI blueprints you won’t find anywhere else

Brief: This quarter's AI transformation blueprint will be released on October 2, 2026, direct to Boardroom members' inboxes. A previous blueprint helped a Fortune 500 CAIO grow AI opportunity pipeline value 42% and cut projected time-to-production 28%.

What’s inside The Boardroom:

• Executive AI transformation blueprints (each quarter)

• The Executive AI Index (349 playbooks)

• 35+ full AI-native professional guides

• Extended version of Enterprise AI Executive

If you're looking to drive enterprise AI P&L impact, join The Boardroom alongside CXOs, VPs and directors before October 2.

MARKET INSIGHT

Enterprise AI spend takes priority

Brief: BCG found IT leaders have moved past two years of caution, planning far more confident 2026 budgets while funnelling spending decisively toward AI. Its IT Spending Pulse surveyed 423 buyers across North America and Europe.

Breakdown:

• IT buyers expect 2026 budgets to grow 5.8% year on year, up from a planned 3.6% two quarters earlier and above pre-tariff levels.

• Net spending falls in most categories, yet 66% expect to raise AI and ML budgets, with cloud/security enablers AI depends on also gaining.

• CRM and ERP both turned negative as companies prioritise AI over systems of record, and IT services swung down a sharp 19 points.

• Buyers now plan to consolidate suppliers in every category but one: a net 34% intend to expand their roster of AI and ML providers.

Why it's important: The important story is not how much buyers are spending but how narrowly. They are concentrating GenAI and agents on a smaller set of high-impact use cases, and that discipline is showing in AI returns: weighted average ROI has risen to 13.8%, up from 11.2% in BCG's mid-2025 survey.

MARKET & BEST PRACTICE INSIGHT

The reprioritization of IT services

Brief: Arthur D. Little detailed how enterprise IT services firms have lost more than a third of their value since 2021, not because the market is shrinking but because it is reprioritizing what clients buy and how work gets delivered.

Breakdown:

• Client demand is concentrating in five AI growth areas, from application modernisation and data governance to security and copilots.

• The pyramid model of large teams doing low-value coding and testing is being upended, with new governance and evaluation roles emerging.

• Pricing is increasingly moving from time/materials toward outcome/consumption models as clients press for greater transparency.

• Impact and potential strategies differ by archetype, with Indian heritage providers most exposed to pricing pressure (image above).

Why it’s important: AI services spending is set to grow 18.2% a year to $478 billion by 2028, yet valuations still reflect the post-pandemic transformation slump. That gap gives private equity an opening to acquire underpriced assets, and Arthur D. Little sets out what to look for.

AI-NATIVE PROFESSIONAL

How to create presentations with ChatGPT Work

Brief: In this guide, you'll learn how to create or revise a PowerPoint or Google Slides presentation with ChatGPT, working from your own source material, an existing reference deck, or a reusable template you can apply to future decks.

Step-by-step:

1. Start with materials you have: notes, a memo, research, or an existing deck. Tell ChatGPT who it's for and what should stay unchanged.

2. In the ChatGPT desktop app, use @Presentations to build a deck from source material. Inside an open file, use the PowerPoint add-in.

3. ChatGPT drafts a short slide plan, builds the presentation, and checks the rendered slides, preserving approved figures and your branding.

4. Use follow-up prompts to match a reference, adapt the story, bring in new information, or save a format you expect to use again.

Best practice: Template Creator saves a personal template backed by the original presentation, so you can reuse the same format in future tasks.

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

BEST PRACTICE INSIGHT

The harness is not enough

Brief: Cognizant shared how an agent harness (orchestration, context etc.) that makes an agent deployable cannot make it reliable across complex tasks or economical at scale. Four capabilities beyond it decide if agents can be trusted.

Breakdown:

• Private, task-level evals against your own environment are what prove an agent's output is correct, and public leaderboards cannot.

• Most deployed agents are static; a closed loop feeding live traces and evals back into training is what compounds value over time.

• Many failures reflect an agent misreading enterprise meaning; a knowledge graph or business rules give structure to reason against.

• A text-only LLM judge can only tell whether output looks right; reactive verification checks formulas, sources and system changes.

Why it’s important: A strong harness was the right place to start, but it is not enough. Measuring agents against their own work, grounding agents in meaning, closing the learning loop and verifying whether the work is actually right turns agentic AI from a cost center into a durable, appreciating asset.

Salesforce released a step-by-step guide to scaling agentic AI and published a 45-page report on questions marketing leaders face.

McKinsey published its 143-page Technology Trends Outlook 2026, covering 14 trends from AI infrastructure to model architectures.

BCG outlined five AI lessons CEOs can take from Asia-Pacific, and explored harness engineering as the operating system for agents.

Bessemer Venture Partners outlined five trends reshaping the C-suite and how CEOs can find executives best equipped for what's next.

McKinsey explored the agentic transformation office and how agentic AI can cut the coordination tax that slows workflows.

Bain interviewed IBM CEO Arvind Krishna, who argued the era of AI experimentation is over and investment should narrow to a few bets.

OpenAI introduced Astra for Law, combining GPT-6 Astra, its most powerful model, with tools and context tailored for legal work.

Accenture and Anthropic partnered on independent evaluation of frontier AI, both investing at least $1 billion over five years.

Anthropic rolled out a beta Projects overhaul letting Claude split work across coding sessions that outlive logout, and is reportedly weighing a pre-IPO model.

Jensen Huang, Elon Musk and Mark Zuckerberg reportedly convinced Trump last month to block a new industry-funded AI oversight body.

Anthropic merged Cowork and Chat into one Claude app, and launched Claude Docs and Slides, its Word and PowerPoint rivals, in beta.

Z AI revealed that GLM-5.3 helped build the 100K-chip cluster now serving GLM-5.3-Flash, and said AI systems are "our successors."

CAREER OPPORTUNITIES

Anthropic - Head of GTM Strategy

Jeffries - AI Strategy Vice President

AstraZeneca - Head of Enterprise AI

EVENTS

Anthropic - Claude for Customers - September 23-25, 2026

HFS - AI Change Roundtable - December 2, 2026

AWS - re:Invent - November 30-December 4, 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.