Generative AI Enterprise

Get the top 1% of generative and agentic AI for enterprises twice a week in under five minutes. Join thousands of executives and professionals. 

Best practices, case studies and innovation insights.

Get on the list: https://generativeaienterprise.ai/subscribe 

Advisory currently at capacity, but do reach out for future requirements.

Writer - Strategist & Architect

Stealth Startup

Founder

Deloitte

Progressed to lead on various AI engagements and initiatives

Generative AI Lead

Goldman Sachs

Summer Analyst

Helping you level up with Generative AI. Follow for daily insights. Deloitte AI, Goldman Sachs

Hi, Lewis here! How can I help you?

Get the top 1% of generative and agentic AI case studies, best practices and innovation insights, twice a week in under five minutes. Join thousands of executives and professionals:

https://generativeaienterprise.ai/subscribe

---

I help executives and professionals unlock the full potential of generative and agentic AI for themselves and their enterprises.

Leveling up with generative and agentic AI can be difficult, but it doesn’t have to be.

With the right guidance and resources, growth is not just a process of trial and error.

Finding solid professional advice to help you level up fast with generative and agentic AI can be tough.

That’s where I come in.

I've been there...

I struggled when I first joined Deloitte AI.
I struggled when I founded my first company.
I struggled when I started mastering generative AI.

But I leveled up, helping executives and dozens of enterprises transform with generative, agentic AI, and broader AI technologies.

Shaping strategies, redesigning operating models, and deploying AI solutions at scale to drive enterprise transformation.

And now I want to help you and your enterprise do that too.

Are you ready to start? Are you truly committed to unlocking your full potential?

Sign up for my twice-weekly briefing, Generative AI Enterprise.

We cover everything from advancing your professional growth with proven case studies to best practice insights and transforming your enterprise through agentic applications.

We’ve got all your bases covered.

We professionals feel fulfilled when we’re growing and improving.

And we can go further and faster with generative AI.

I want that for you.

Take the first step today.

Join us here: https://generativeaienterprise.ai/subscribe

I reviewed 147 AI playbooks from the past 6 months.
These 18 are essential reads for AI leaders:

Leaders across tech, finance, sales, marketing, HR, risk, and beyond are racing to unlock value from AI agents, agentic AI, and generative AI.

Strategies, operating models, architectures, delivery approaches, and responsible AI practices are evolving rapidly.

I’ve distilled the key takeaways from 18 top AI leadership playbooks published in the past six months to help you save time, and stay ahead.

1. McKinsey’s CEO agentic AI playbook 
2. OpenAI’s 15-page AI leadership guide
3. Google’s agentic services AI playbook
4. BCG's playbook on AI-first enterprises
5. AWS’s 17-page agentic AI executive guide
6. Microsoft’s 40-slide CIO playbook
7. Salesforce’s agentic enterprise playbook
8. Utah’s 98-page AI leadership blueprint 
9. Capgemini 25-page AI transformation blueprint
10. Stanford’s 14-page enterprise AI playbook
11. OpenAI’s AI use case identification guide
12. Deloitte’s board governance playbook
13. BCG’s 37-slides on enterprise AI agents
14. WRITER’s CMO on AI reshaping marketing
15. PwC Strategy&’s 20-page AI adoption approach
16. Palo Alto Networks AI agent defense strategies
17. Teneo published four CEO software playbooks
18. BCG's eight ~25-slide playbooks for CxOs

See the image below for takeaways and full playbooks.

P.S. For more like this: https://lnkd.in/e4iRMQiY

P.P.S. I will get to everyone, thanks for your patience.

Lewis Walker ➲

Get the top 1% of agentic/gen AI (View my newsletter)

McKinsey just shared 6 agentic AI lessons:
(from 50+ builds they've led)

Their just released insights span six key lessons learned (L1-6) that enterprises can apply to help capture value from agentic AI.

→ It’s not about the agent itself; it’s about the workflow (L1). The biggest gains come from reimagining people, processes and technology.

→ Workflow focus enables teams to apply the right technology at the right point (L2), particularly important in complex/multi-step workflows (page 4).

→ Invest in evaluations (L3) to assess agent performance and improve outputs, and make it easy to track and verify each workflow step (L4).

→ Develop agents and agent components that can be reused across workflows (L5), helping reduce effort by 30 to 50 percent on average.

→ Humans remain essential, though roles evolve (L6). People need to oversee accuracy, exercise judgment, and handle edge cases.

Full insights detailed in the 12 pages below.

For more agentic AI lessons: https://lnkd.in/eBBHY3H7

New: McKinsey Lessons 50+ Agentic AI Builds

Want to prioritize AI use cases faster?


Enterprises often struggle to identify AI use cases with the highest potential and prioritize them to drive maximum impact.

The wide range of agentic AI, AI agents, and generative AI applications is a blessing and curse, which can make it hard to know where to invest.

AI use cases need to be assessed for feasibility and value in the context of enterprises’ strategic objectives and nuanced risk profiles.

Existing frameworks from top firms can help accelerate the process.

Here are 12 examples:

1\ BCG – Outlines how BCG assesses gen AI opportunities (slides 6-12), factors affecting task automatability, and cost reduction potential.
2\ Google – Uses a matrix to plot potential AI use cases based on expected value, actionability, and feasibility to guide prioritization decisions.
3\ OpenAI – Impact/Effort method (page 27) details how customer success teams prioritize AI use cases for enterprises efficiently and effectively.
4\ Capgemini – Measures how amenable areas are to generative AI and identifies roadblocks or gaps in adoption to guide delivery (page 14).
5\ PwC – Evaluates AI use cases for value and feasibility (page 8) using industry benchmarks, cross-industry experiences, and more. 
6\ Anthropic – Shares how AI teams select optimal use cases (page 16), leveraging data, success metrics, expected ROI, and more. 
7\ Gartner – AI Opportunity Radar enables enterprises to map AI ambitions, helping executives explore, prioritize, and commit to use case opportunities.
8\ Microsoft – Shows how CIOs link cost control to value narratives (page 23), drive transformation and earn broader executive buy-in.
9\ WRITER – Agentic AI ROI is a four-part model measuring the real business impact of applications across use cases. 
10\ KPMG – AI investment assessment, factoring hardware, software, manpower, and projected outcomes to guide decision-making.
11\ IDC – Prioritize AI use cases and build an opportunity portfolio (page 3) based on business value and feasibility for quick wins and strategic bets.
12\ Scan AI – A 7-step process identifies agentic AI use cases, aligning with core value streams, persona-centric design, and more.

See the graphic below to get the full frameworks.

P.S. For more like this: https://lnkd.in/eBBHY3H7

Want 2195 enterprise AI use cases? 
(fast-track value below)

The process of finding the right AI use cases can vary by enterprise context and maturity, but certain approaches have been proven to deliver results.

Here’s a simple one (leveraged by OpenAI):

Empower employees to start with high impact, low effort quick wins to build momentum.

It sounds obvious, but many enterprises overlook this and dive into complex use cases too early, stalling progress.

1\ Understand where AI adds value

Ask your teams to list scenarios and tasks where they:

→ Struggle to get started or run into blockers.
→ Spend a lot of time on manual work that others don’t fully appreciate, or that isn’t the best use of their time (i.e. their “anti to-do list”).
→ Hit a skills bottleneck until another team can help (e.g. data analysis, design, on-brand writing).

2\ Teach fundamental use cases to fast track value

→ Educate your teams on use cases that typically work and provide department-specific examples (2195 in the graphic below)
→ Next, start brainstorming use case ideas, and running hackathons to source the most impactful ones.
→ Set up a centralized location to collect all the use cases your teams come up with.

3\ Prioritize use cases

→ Promote the prioritization framework company-wide, encouraging employees to use it in team meetings to identify the best ideas.
→ Have leaders champion use cases that impact entire departments. Top-down support is critical for successful AI deployment.
→ Re-evaluate use case scoring each quarter, as today’s high-effort tasks may become low-effort as AI capabilities advance.

Want to fast-track value with AI? 

Tried and tested AI agent, agentic AI, and generative AI use cases from companies such as OpenAI, Google, McKinsey, Microsoft, Deloitte can help.

See the image below for more details.

The trillion-dollar AI software development stack:
(according to Andreessen Horowitz)

New insights from Silicon Valley VC firm Andreessen Horowitz (a16z) explore the latest on how AI is changing software development.

→ a16z estimates AI software development could drive $3T in annual economic impact, roughly equal to the GDP of France.

→ What will AI software development look like? a16z maps what it's seeing today (top image), showing where startups are building (orange boxes).

→ More categories appear in the full market map (bottom image). Alongside startups, products from incumbents are occasionally included.

a16z’s full insights dive deeper: https://lnkd.in/eFwcA8Mb

For more: https://lnkd.in/e4iRMQiY

McKinsey’s 2-year roadmap for agentic AI is out:
(page 13 for key CEO actions)

New insights from McKinsey outline how CEOs can navigate early growing pains and consider the right strategic decisions over a two-year timeline.

1\ Many executives still have limited notions of what agents can do. McKinsey defines them by task complexity: augmentation to x-function (p.5).

2\ In year one momentum builds as fluency grows. Agents start automating work with 1st-gen AI tech, shifting productivity and some roles (p.7).

3\ Year two focuses on scaling impact. Measures shift from activity metrics to P&L impact as agent-to-FTE ratios rise across business functions (p.10).

4\ CEOs and boards need to operate at two speeds by driving short term transformation while thinking longer-term (see five CEO actions p.13).

The shroud of uncertainty that still veils AI agents shouldn’t hold leaders back. In fact, this “trough of disillusionment” is a chance for decisive executives to outpace competitors, leveraging thoughtful, strategic action to unlock new opportunities for productivity and enterprise growth.

Full goals, decisions, and implications for CEOs below.

P.S. For 18 more executive AI playbooks: https://lnkd.in/e5VBUzsB

New McKinsey Executive AI Insights

How to become an AI-first enterprise:

OpenAI’s 18 steps span: 

How to align employees, leadership on AI strategy.
How to enable and motivate teams to use AI.
How to amplify AI wins and use cases across teams.
How to speed up decisions to keep pace with AI.
How to run lightweight guidelines for responsible AI.

Master the 18 steps below to become AI-first.

👉 Save this to revisit later.

P.S. Link to full guide in comments.

Boston Consulting Group analyzed 4 billion FTE hours to quantify AI impact: 

Leveraging its proprietary GenAI Workforce Diagnostic tool, the analysis spans seven functions (customer service, comms, HR etc.) 

At the function level (slides 2–3), it evaluates productivity gains (potential time reduction %) and cost reduction potential (%) resulting from: 

1\ Task automation/augmentation w/ NLP, RPA, GenAI
2\ Process reshaping/optimization with agentic AI

The analysis also drills down to sub-function and task level activities and associated benefits (slides 4–5).

BCG’s methodology is outlined slide 7 onwards, inc. factors influencing task automatability (slide 10). 

P.S. Get 11 more methodologies for prioritizing AI opportunities (free): https://lnkd.in/eMnxN5ch

BCG Analyzes 4B FTE Hours 

Want to build AI agents from scratch?


Salvatore Raieli and Gabriele Iuculano published an extensive, practical guide to Building AI Agents with LLMs, RAG, and Knowledge Graphs

Unlike many overly simplified or fragmented books, this guide combines strong theory with hands-on implementations for production AI agents.

Reviewers include Malhar Deshpande and Lalit Chourey from Meta.

Structured in three parts:

1\ The AI Agent Engine: From Text to LLMs

→ Lays the foundations for how modern AI agents handle language, starting with representing raw text as numbers for deep learning models.

→ Shifts to the Transformer model and explains how attention mechanisms revolutionized natural language processing.

→ Examines how LLMs are built by scaling transformers, discussing training, instruction tuning, fine-tuning, and general-purpose reasoning. 


2\ AI Agents and Retrieval of Knowledge

→ Focuses on extending the capabilities of LLMs by enabling them to access, retrieve, and reason over external knowledge sources.

→ Explores the creation of agents that interact with the web, retrieve live information, and perform tasks beyond question answering.

→ Addresses RAG, progressing from basic pipelines to modular, scalable systems that reduce hallucinations and improve accuracy.

→ Structured knowledge via knowledge graphs (GraphRAG) is introduced as a powerful method to represent and reason over information.

→ Examines how reinforcement learning can be used to align agent behavior and improve decision-making


3\ Creating Sophisticated AI to Solve Complex Scenarios

→ Focuses on assembling the components in the previous chapters to build fully-fledged, production-ready AI systems.

→ Covers the design and orchestration of single- and multi-agent systems where LLMs collaborate with tools, APIs, and other models.

→ Practical guidance follows on building and deploying AI agents, including asynchronous programming and containerization.

→ Concludes on the future of AI agents, their industry impact, and the ethical and technical challenges that lie ahead.

Build AI agents from scratch: https://lnkd.in/eZAVpKFU

Building AI Agents From Scratch

Want to build production AI agents?


AI Agents in Practice, 282 pages published by Microsoft Architect Valentina Alto, shows how to design, implement, and scale production AI agents.

It covers AI agent architecture and real use cases, giving developers, architects, PMs, or AI enthusiasts the practical skills to leverage AI agents.

Structured into three parts:

1\ Foundations of AI Workflows and the Rise of AI Agents (Chapters 1-2)

→ AI workflows since the rise of generative models, evolving from simple API calls to more intelligent, autonomous behaviors.

→ AI agents, their LLMs, tools, memory, and context, and highlights the growing need for agentic systems across industries.

2\ Designing, Building, and Scaling AI Agents (Chapters 3-7) 

→ Practical aspects of agent development, including orchestration tools, memory and context handling, tool integration, and agent observability. 

→ Building single-agent and multi-agent apps with LangChain and LangGraph, using hands-on examples like customer support agents.

3\ Road to an Open, Agentic Ecosystem (Chapters 8-9)

→ Protocols, platforms, and principles shaping the future of intelligent software, including open standards like MCP, A2A, and NLWeb.

→ Building responsible, secure, and cost-effective agent systems for enterprise-scale deployment.

Hands-on examples and code included.

Check it out here: https://lnkd.in/e4yxUkR5

Design, Implement & Scale AI Agents

Deloitte on finding the right agentic AI use cases:
(page 9 for prioritization matrix)

1\ Qualify use cases for agentic AI suitability

→ Organizations begin with specific business problems and seek to determine whether agentic AI is the right solution. 
→ The focus is on determining whether the use case requires autonomy, contextual awareness, and adaptive decision-making.

Deloitte’s criteria to evaluate if a use case is agentic:

↳ Reasoning, complexity, context dependence
↳ Autonomy with escalation when needed
↳ Clear logical end
↳ Action- and goal-driven workflows
↳ Multistep, interconnected tasks
↳ Cyclic, repetitive processes
↳ Non-explanatory workflows
↳ Potential for continuous learning & optimization

See page 7 for more details and examples of each.

2\ Prioritize use cases

→ Once a portfolio of use cases exists, the enterprise's challenge shifts to prioritizing and building a roadmap for agentic AI implementation.
→ A prioritisation framework (see page 9) helps assess each use case by impact, implementation effort & differentiability.
→ The framework includes the "Differentiability Index" to measure how essential agentic AI is for a particular use case.
→ Overall, the framework guides which use cases should be agenticised first, and why.

P.S. Get 12 more AI use case prioritization frameworks: https://lnkd.in/eMnxN5ch

New Deloitte Agentic AI Best Practices

OpenAI’s first-of-its-kind enterprise usage analysis.
Key insights:

The new analysis, leveraging data directly from ChatGPT on enterprise usage/adoption, includes addressing two key questions:

1\ Which industries are U.S. ChatGPT weekly active users (WAU) in?

→ Usage is highest in IT (27%), professional, scientific & tech services (17%), and manufacturing (10%) (p.9).
→ Healthcare adoption lags at 5%, slowed by privacy rules, compliance burdens, and risk-averse cultures, despite it's size/data intensity.
→ Data comes from U.S. ChatGPT Free, Plus, and Pro plan users with professional emails, mapped to industry categories.

2\  How are departments using ChatGPT Enterprise in their first 90 days?

→ Usage is dominated by writing, research, programming, and analysis.
→ ChatGPT Enterprise plan data was mapped across 14 capabilities (e.g., writing, factual Q&A, coding) in 17 departments (p.12).
→ In Product, for instance, usage breaks down as: 33% writing, 25% factual Q&A, 11% programming, and 8% media generation & analysis.

Further insights below.

View my newsletter for more enterprise AI.

New OpenAI Enterprise Usage Analysis

OpenAI just shared what actually drives AI adoption.
15 practical actions to get started:

OpenAI's new playbook offers five steps: Align, Activate, Amplify, Accelerate, Govern, to help leaders drive AI adoption and build AI-first organizations.

Align

Employees adopt faster when leaders link AI to their skills, purpose, and the company's competitive advantage, while openly showing support.

1\ Run periodic pulse surveys to gauge employee clarity on AI strategy.
2\ Track progress toward AI adoption goals with an open dashboard.
3\ Review frequency/feedback from functional leader-team AI sessions.

Activate

Nearly half of employees say they lack AI training; leaders should encourage experimentation, department-specific learning, and normalize upskilling.

4\ Track daily/weekly active users, GPT shares, training enrollment/completion. Support low-adoption teams.
5\ Add AI to career pathways/reviews. Track how AI impacts promotions.
6\ Monitor how many ideas move past prototype into production.

Amplify

Scaling AI impact requires sharing knowledge, use cases, prompts etc. across teams to turn isolated successes into reusable progress.

7\ Share at least 3 AI wins monthly via newsletters, meetings etc.
8\ Maintain one central hub with a clear owner and update it twice monthly.
9\ Track participation in AI forums: contributors, shared resources etc.

Accelerate

Acceleration depends on removing friction: flexible infra, clear decision-making authority and lightweight approvals to move pilots into prod fast.

10\ Track time-to-production for AI projects. Review blockers monthly.
11\ Review request/approval times for AI tools, platforms, and datasets.
12\ Track projects funded/staffed compared to business impact.

Govern

Moving fast doesn’t mean ignoring risks. Clear, practical governance guidelines support rapid action within established safeguards. 

13\ Use check-ins/project reviews to confirm “safe-to-try” guidelines.
14\ Review governance quarterly with legal, risk, and functional input.
15\ Track project delays from governance. Gather feedback and adjust.

P.S. What would you add?

Full playbook below, and linked in comments, with further guidance.

--- 

📌 Want 30+ AI playbooks for execs/professionals?
Get them here: https://lnkd.in/e4iRMQiY

New OpenAI Leadership Playbook

5 examples of OpenAI using OpenAI in production:


OpenAI launched OpenAI on OpenAI, a series showing how they leverage AI internally in live deployments.

The goal is to share patterns enterprises can adopt, starting with five AI solutions addressing key business challenges:

1\ GTM Assistant: A Slack-based tool streamlines research, meetings, and product Q&A, improving average sales rep productivity by 20%.
↳ https://lnkd.in/edhzsF-F

2\ DocuGPT: Converts contracts into structured, searchable data, helping finance teams review faster, more consistently, and at scale.
↳ https://lnkd.in/e3Wvn3pV

3\ Research Assistant: Turns millions of support tickets into conversational insights, surfacing trends to help act on customer feedback quickly.
↳ https://lnkd.in/ewT2vkHq

4\ Support Agent: Uses AI agents, continuous evals, and dynamic knowledge loops to turn every interaction into training data and improve quality.
↳ https://lnkd.in/eGqyb6Ps

5\ Inbound Sales Assistant: Personalizes lead responses, raising accuracy from 60% to over 98% within weeks, and routes qualified prospects.
↳ https://lnkd.in/eSNdMGY5

For 70+ more proven AI case studies, view my newsletter.

How OpenAI Uses OpenAI

Want to actually become AI-first?


Enterprises often ask: Where is our competition, how can we stay ahead and build an AI-first organization?

More specifically, how can we:

→ Align employees and leadership on AI strategy
→ Activate and motivate teams to use AI
→ Amplify wins and use cases across teams
→ Accelerate decisions to keep pace with AI
→ Govern with lightweight guidelines/responsible AI

Every enterprise will become an AI enterprise. The question is not if, but how fast and how well.

OpenAI’s recent leadership guide includes 18 steps to get started becoming AI-first. 

Key actions are in the PDF below.

Full guide: https://lnkd.in/eBV9fRgc

For more AI-first: https://lnkd.in/eBBHY3H7 

Sawsen Haj Amor 🎙

How to Become an AI-First Enterprise

Video is no longer one-way. 
Synthesia just launched Video Agents:

Video Agents are the first step toward video becoming a true two-way conversation.

These LLM-enabled, lifelike avatars can talk, listen, and act in real time, bringing a level of two-way interactivity that feels like a live conversation. 

Video Agents can:

→ Run training sessions
→ Screen job candidates
→ Guide customers through learning experiences

You can insert them into any point in your video to start a real-time conversation with viewers, and combine them with other formats.

But that’s not all.

The Synthesia 3.0 launch also introduces broader updates across avatars, voices, interactivity, and AI dubbing, plus Veo 3 for enterprise, all already live.

And coming soon to the AI video platform trusted by 90% of the Fortune 100: Video Agents, Copilot, and further advances in avatars and interactivity.

For full details on the future of AI video, see my takeaways below.

Turn to page 7 for Video Agents.

P.S. How will you leverage Video Agents?

The New Era of Video, Synthesia 3.0

The 2025 State of AI Report is out.
5 insights from the 300 slides:

Air Street Capital, the VC firm backing AI-first companies, just released its 8th annual report covering research, industry, politics, and safety.

Example insights from the 2025 report:

1\ Reasoning goes mainstream with OpenAI, DeepMind, Anthropic, and DeepSeek turning “think-then-answer” into real products.

2\ 44% of U.S. firms now pay for AI tools (up from 5% in 2023), average contracts reach $530K, and AI-first startups grow 1.5x faster than peers.

3\ Multi-GW data center initiatives like Stargate signal a new wave of compute infrastructure with power supply emerging as the new constraint.

4\ From DeepMind’s Co-Scientist to Stanford’s Virtual Lab, models are generating, debating, and validating new discoveries.

5\ The U.S. leaned into “America-first AI,” Europe’s AI Act stumbled, and China expanded its open-weight and domestic silicon ambitions.

Report below and here: https://lnkd.in/ew_csCSx

P.S. View my newsletter for more.

Taplio

Lewis Walker ➲'s Linkedin Analytics

's Best Posts (last 30 days)

Want to drive more opportunities from LinkedIn?

Taplio

.css-1jgreyo{background-image:linear-gradient(284deg, rgba(39,43,72,1) 0%, rgba(33,47,183,1) 100%);color:transparent;-webkit-background-clip:text;background-clip:text;}Lewis Walker ➲'s Linkedin Analytics

's Best Posts .css-qqfgvy{font-size:var(--chakra-fontSizes-md);}(last 30 days)

Want to drive more opportunities from LinkedIn?

Lewis Walker ➲'s Linkedin Analytics

's Best Posts (last 30 days)