Jonny Tooze's Recent LinkedIn Posts

Claude Mythos is real. And it leaked before launch. Anthropic's most powerful model ever was not announced at a keynote. It was discovered in a misconfigured content management system. Nearly 3,000 internal documents left publicly searchable because of a default setting nobody changed. The irony writes itself. The company building the AI that finds software vulnerabilities just got caught by one. But the model itself is the real story. ⚡ Here is what the leaked documents reveal: 🔹 Dramatically higher scores than Claude Opus on coding, reasoning, and cybersecurity tasks 🔹 Described internally as "a step change" in capabilities 🔹 Already being piloted with early-access customers 🔹 Internal warnings that it could "significantly heighten cybersecurity risks" by rapidly finding and exploiting software vulnerabilities This is a huge jump from AI-assisted to AI-agentic at a level we have not seen. 📊 For context: I write production code with Claude every day. The current Opus model already changed how I build software. A step change above that is not a product update. That is a new category. Here is what this means for engineering leaders: 1️⃣ The capability ceiling just moved again. Whatever you planned your AI strategy around six months ago is already behind. 2️⃣ The cybersecurity implications are real. A model that finds vulnerabilities faster than humans can patch them changes the threat landscape for every organisation. 3️⃣ The rollout will be slow and controlled. Anthropic is doing a security-first release. If you are not already in their early-access programme, you will be waiting. The companies that redesigned their delivery models around the current generation of AI are about to get a massive compound advantage. The ones still running pilots are about to fall further behind. This is not a moment to wait and see. This is a moment to be ready. 👉 Follow Jonny Tooze for more on what this means for AI-native engineering. ♻️ Repost if your CTO needs to see this before Monday.

You don't need to be an engineer to lead an AI strategy this year. You just need the right vocabulary. It is easy to get lost in the jargon. But if you just nod along when your product team talks about "Parameters" and "Chain-of-Thought," it becomes nearly impossible to manage your roadmap. Fluency is a leadership skill. If you confuse "Generative AI" with "Machine Learning," it is easy to misallocate your budget. To help clear the noise, I put together a plain-English guide of the 85 essential terms every executive needs to understand. Save this cheat sheet for your next strategy meeting: 🟢 The Foundations: AI vs. Machine Learning vs. Deep Learning. 🟢 The Systems: AI Agents, Tool Use, and Chain-of-Thought. 🟢 The Architecture: Transformers, Parameters, and Context Windows. 🟢 The Risks: Bias, Overfitting, and Prompt Injection. (Zoom in on the image to read the plain-English definitions). Fluency is the first step to execution. Want to stay ahead of the curve? We send weekly insights on building AI-native organizations. Subscribe free 👉 https://lnkd.in/eJnCpjMQ ♻️ Repost this to get your whole team on the same page.

The 10 Anthropic Playbooks every engineering leader needs to read this week. If you are leading an AI transformation this year, bookmark this post. Most leaders are struggling to leverage Claude Code because they lack a clear deployment framework. Anthropic just eliminated that excuse by dropping their official enterprise playbooks. The 10 mandatory reads for your leadership team: 1. Anthropic Economic Index - https://lnkd.in/eavNSCcd 2. 2026 Agentic Coding Trends - https://lnkd.in/eiabhEKH 3. Scaling Agentic Coding (The deployment manual) - https://lnkd.in/eumwZu9c 4. How Anthropic Teams Use Claude - https://lnkd.in/eg87C2ai 5. The Code Modernization Playbook (Legacy tech debt) - https://lnkd.in/eKXMfqF2 6. Building Skills for Claude - https://lnkd.in/en9k-JSg 7. Measuring Agent Autonomy (Governance & Safety) - https://lnkd.in/es4bpGqh 8. Impact on Coding Skills (Junior vs Senior mastery) - https://lnkd.in/eS965gxS 9. Equipping Agents for the Real World - https://lnkd.in/eaqAy5eB 10. Introducing Cowork (AI for non-technical teams) - https://lnkd.in/e4yY5WJ2 Frameworks are useless without execution. If you need help building the organizational structure to support these playbooks, we dive deep into this inside our ENDGAME AI-Native Leadership Cohort. 👉 Take a look here: https://lnkd.in/e4qKrm_E Follow Jonny Tooze for more straight-talk on AI, product, and delivery.

Stop treating "AI" like a single checkbox on your strategy roadmap. AI isn't a single technology. It is an evolutionary four-part maturity model. If your company is still just predicting outcomes or writing faster emails, you are falling behind the market. Here is how to explain the four types of AI to your leadership team: 1. Predictive AI (The Analyst) What it does: Forecasts customer behavior and spots anomalies using historical data. The Reality: This is traditional data science. In 2026, it is table stakes. 2. Generative AI (The Creator) What it does: Generates content, code, and conversational chat bots from prompts. The Reality: This is where 90% of companies are stuck right now. High augmentation, but limited automation. 3. AI Agents (The Doer) What it does: Steps out of the chatbox to connect to APIs and complete standalone tasks like resolving tickets. The Reality: You finally move from "chatting" to "executing." 4. Agentic AI (The Workforce) What it does: Chains multiple steps into end-to-end automated processes using multi-agent orchestration. The Reality: This is the finish line. You aren't just speeding up a human task; you are delegating entire workflows to build AI-native products. Speed is your ultimate advantage in the era of AI. If you don't know how to move your product org from Level 2 to Level 4, we spend 8 weeks teaching this exact transition inside our ENDGAME AI-Native Leadership Cohort. 👉 Apply here: https://lnkd.in/e4qKrm_E Follow Jonny Tooze for more straight-talk on AI, product, and delivery.

You don't have time to read every AI report published this year, so we filtered them for you. If you are leading a digital transformation in 2026, this is your required reading list. We've cut through the noise and gathered the 10 most critical strategy playbooks from the top firms in the world. From OpenAI's real-world deployment data to IBM's operating models, this is your blueprint for the year. Bookmark this for your next strategy session: 1/ OpenAI: Enterprise AI Data - https://lnkd.in/ew9tYfxy 2/ Anthropic: 2026 AI Agents - https://lnkd.in/eCn2g_Nf 3/ McKinsey: State of AI - https://lnkd.in/eWWH8ExC 4/ Accenture: Platform Strategy - https://lnkd.in/eqrcHHnS 5/ BCG: Leading the Agent Era - https://lnkd.in/eb4Ujpg4 6/ EY: Top 10 ROI Opportunities - https://lnkd.in/e6fBP2UZ 7/ Bain: The Agentic Foundation - https://lnkd.in/exs82CEs 8/ IBM: Operating Models - https://lnkd.in/e2cVZjqT 9/ McKinsey: The Agentic Org - https://lnkd.in/egM_cbfc 10/ ENDGAME: Product Blueprint - https://end.game/ Reading the blueprint is step one. Building the house is step two. If you need the architectural support to actually implement these frameworks across your org, that is exactly what we will teach you in our ENDGAME Leadership Cohort. We’ll help you turn these PDFs into production workflows. 👉 Apply here: https://lnkd.in/e4qKrm_E

The 60-Second Executive Cheat Sheet for Claude Cowork. You probably aren't ready to navigate the newest AI platform blind. So I built a one-page blueprint covering everything you need to know. People are compressing full-day workflows into minutes. Here is how: 1. Understand the Capability Cowork is an AI assistant that lives on your desktop. It reads files, controls your browser (Salesforce, Notion, Slack), and produces finished PPTs, PDFs, and Spreadsheets. 2. Know the Difference Claude Code: For engineers. Lives in the terminal to build software. Claude Cowork: For business teams. No IT ticket required. Designed to run operations. 3. Deploy in 24 Hours Download the app, install the Chrome extension, mount your folders, and install a department plugin (like Sales or Finance). Start with one repeatable task. Prove the value. Then scale. (Zoom in on the image for the exact prompts your CFO, CLO, and VP of Sales should be using today). Ready to build an AI-Native organization? We teach the deployment frameworks in our next ENDGAME Leadership Cohort. 👉 Apply here: https://lnkd.in/e4qKrm_E ♻️ Repost to help your network master Cowork. ___________________________________________________________ Thanks to Alex Barády for the spark behind this one. Follow Jonny Tooze for more straight-talk on AI, product, and delivery.

You can't manage a roadmap if you can't define the roadmap. If you just nod along when your engineering team talks about "Agentic Tool Use" or "Context Windows," you are surrendering control of your strategy. You don't need to write Python to lead a company in 2026. But you absolutely must be fluent in the language of the people building your products. If you confuse "Machine Learning" with "Generative AI," you are going to misallocate your budget. We mapped out the 85 essential terms every non-technical executive needs to understand this year. Keep this cheat sheet on your desktop to decode the noise: 🟢 The Foundations: Stop confusing AI, Machine Learning, and Deep Learning. 🟢 The Systems: Understand how AI Agents, Tool Use, and Chain-of-Thought actually execute tasks. 🟢 The Architecture: Decode the infrastructure jargon like Transformers and Parameters. 🟢 The Liabilities: Protect your P&L by understanding Bias, Overfitting, and Prompt Injection. (Zoom in on the image to read the plain-English definitions). Fluency is the first step to execution. Stop translating and start leading. Want to stay ahead of the curve? We send weekly insights on building AI-native organizations. Subscribe free 👉 https://lnkd.in/eJnCpjMQ ♻️ Repost this to get your whole team on the same page. ___________________________________________________________ Follow ENDGAME if you care about faster delivery, higher adoption, and measurable returns from AI – not dashboards full of vanity metrics. Thanks to Alex Barády for the spark behind this one. Follow Jonny Tooze for more straight-talk on AI, product, and delivery.

Why can your ex-CTO build an app in a weekend, while your team takes three weeks to ship a feature? Every executive is asking the same question right now: "If AI coding is so powerful, why isn't our company moving any faster?" The timeline on this board explains exactly where your bottleneck is. Phase 1: The Individual (Solved in Nov 2025) We already solved Agentic Coding for the solo developer. That is why you see individuals achieving impossible output. Phase 2: The Team (The Current Battlefield) This is where 90% of companies are stuck right now. You cannot just hand a solo tool to a 10-person team and expect a 10x return. To make Agentic Coding work at the team level, you have to completely break and rebuild your existing processes, roles, and tooling. Stop optimizing for a 25% gain. Redesign the team for a 250% gain. Phase 3: The Organization (Arriving 2027) Org-level transformation is coming faster than you think. But you cannot scale this to the whole company until you prove it inside a single squad. Stop expecting 2027 organizational results when you haven't even solved the 2026 team workflows. Need the blueprint for Phase 2? We are helping dozens of enterprises figure out how to scale Agentic workflows at the team level right now. Join our AI-Native Leadership Cohort to get the playbook. 👉 Apply here: https://lnkd.in/e4qKrm_E

Stop trying to invent your AI strategy from a blank page. Every week, I speak to executives who are paralyzed. They are trying to figure out their AI operating model and tech stack entirely from scratch. They are wasting time. The heaviest hitters in the industry have already open-sourced their blueprints. I compiled the definitive reading list for 2026 into one whiteboard. Notice how it is split into two distinct disciplines: 1. The Strategy Playbooks This is for your business leaders. Before you write a single line of code, you need to understand the operating model. These frameworks from McKinsey, BCG, and IBM teach you how to align your org chart, identify use cases, and manage risk. 2. The Architecture Blueprints This is for your engineering leaders. Once the strategy is set, you need the tech stack. AWS, Google, NVIDIA, and Databricks have published the exact reference architectures required to run Generative AI safely at an enterprise scale. Stop guessing. Read the blueprints. Reading the playbook is step one. Building the organization is step two. If you need the architectural support to actually turn these PDFs into production workflows, we have a few spots left in our next ENDGAME Leadership Cohort. 👉 Apply here: https://lnkd.in/e4qKrm_E Follow Jonny Tooze for more straight-talk on AI, product, and delivery.

The grace period for Enterprise AI is officially over. Boards are looking at their software budgets and asking CTOs a very uncomfortable question: "We bought the tools. Where is the 2.5x speed increase?" If you are an executive fielding that question right now, look at the visual below. You are likely stuck in the crowd at door number two. The McKinsey Reality Check: 88% of companies bought AI tools. Only 6% actually redesigned their organization. Here is why the 88% are failing: -They are treating AI like an expensive spell-checker. -They kept the exact same bloated agile rituals, the same spec-to-ticket pipelines, and the same manual QA bottlenecks. -They just gave their developers an autocomplete feature and expected magic. Here is why the 6% are winning: -They didn't optimize the old process; they replaced it. -They moved to Agentic Development and chose to fundamentally Redesign the Organisation. If you don't change how your teams work, the tools are just a very expensive distraction. Need the blueprint to walk through that final door? We spend 8 weeks teaching executives this exact paradigm shift in the ENDGAME Leadership Cohort. 👉 Apply here: https://lnkd.in/e4qKrm_E Follow Jonny Tooze for more straight-talk on AI, product, and delivery.

The 2026 AI Cheat Sheet for the C-Suite. Agentic AI isn’t one single tool. Think of it as a 5-layer stack. Each layer unlocks a completely different business case. If you are an executive, save this breakdown to make sure you are building the right roadmap: 1. Machine Learning (Predictions) Use Case: Forecast demand and detect fraud. Tools: AWS SageMaker, Vertex AI. 2. Neural Networks (Perception) Use Case: Process unstructured documents, voice, and images. Tools: TensorFlow, PyTorch. 3. Generative AI (Creation) Use Case: Draft code, emails, and marketing content in minutes. Tools: Claude, ChatGPT, Gemini. 4. AI Agents (Execution) Use Case: Complete standalone tasks like resolving a customer support ticket from start to finish. Tools: LangChain, CrewAI. 5. Agentic AI (Orchestration) Use Case: Networks of autonomous agents collaborating to modernize legacy software or run entire business processes. Tools: Claude Code, OpenAI Codex. The leap from Layer 3 (Creation) to Layer 5 (Orchestration) is the most important shift of the decade. Want to learn how to lead that transition? Join our AI-Native Leadership Cohort. 👉 Apply here: https://lnkd.in/e4qKrm_E ♻️ Repost to get your leadership team speaking the same language.

The grace period for Enterprise AI is over. Boards are asking CTOs one question this quarter: "We funded the AI tools. Where is the 2.5x speed?" If you don't have an answer, look at this McKinsey data: 88% of companies bought the tools. Only 6% redesigned their organization. ❌ The 88% Mistake: You kept the same bloated agile rituals, manual QA, and legacy pipelines. You just gave your engineers an autocomplete tool and expected miracles. ✅ The 6% Advantage: You didn't optimize the old process; you replaced it. You moved to Agentic Development and completely redesigned how software is shipped. Adding AI to a slow process just gives you a very expensive slow process. Change how you work, or get left behind. Ready to redesign your organization? Learn the exact execution frameworks in our 8-week ENDGAME Leadership Cohort. Join our AI-Native Leadership Cohort 👉 Apply here:https://lnkd.in/e4qKrm_E ♻️ Repost this to fix the communication gap in your company. Follow Jonny Tooze for more straight talk on AI-native delivery.

Most teams think AI agents are just better chatbots. That is like thinking a self-driving car is just cruise control. One follows instructions. The other makes decisions. Here is the difference that matters for your delivery team: 🤖 **AI Assistants (what you are probably using)** You write a prompt. You get a response. You decide what to do with it. The AI waits for your next instruction. Every action requires a human in the loop. The AI is fast, but your workflow is still sequential. You are the bottleneck. ⚡ **AI Agents (what the fastest teams have moved to)** You define the goal. The agent breaks it into steps. It plans, executes, tests, and iterates. It calls tools, reads files, writes code, and asks for help only when it is stuck. The human reviews the output, not every intermediate step. The workflow is parallel. The bottleneck disappears. 📊 The numbers speak for themselves: 🔹 AI Assistants: 25-30% productivity improvement 🔹 AI Agents: 2-2.5x faster delivery on complex work That is not an incremental upgrade. That is a different operating model. Most engineering leaders I speak to are still optimising for assistants. Better prompts. Better completions. Better autocomplete. The teams pulling ahead skipped that entirely. They went straight to agents that own entire tasks end to end. The question is not "are your developers using AI?" It is "are your developers still the ones doing all the work?" If the answer is yes, you are running the old model with a new paintjob. 👉 Follow Jonny Tooze for more straight talk on AI-native delivery. ♻️ Repost if an engineering leader in your network needs to understand this distinction.

The definition of "AI Coding" just changed overnight. If you missed it, your competitors didn't. When Claude Code dropped, it wiped $285 Billion off the market, that wasn't a glitch. That was investors realizing that the era of the "AI Assistant" is already over. We have entered the era of the "AI Agent." If you are an executive, you need to be able to explain the difference to your leadership team today. Use this 3-part framework: Level 1: Vibe Coding (Idea Generation) What it is: Prompt-driven prototyping (Bolt, Lovable). The Value: Speed over precision. It lets business teams build functional mockups in minutes. Level 2: AI-Assisted (Live Support) What it is: Real-time pair programming (Cursor, Copilot). The Value: It makes your developers 25% faster. The human steers, the AI suggests. Level 3: Agentic Coding (End-to-End Autonomy) What it is: Autonomous execution (Claude Code, Devin). The Value: You stop asking for lines of code and start asking for outcomes (e.g., "Migrate this legacy system"). The AI runs a self-correcting loop until the job is done. Right now, 90% of companies are stuck at Level 2, celebrating their 25% productivity bumps. Meanwhile, the smartest teams are preparing for the shift to Level 3. As Andrej Karpathy noted, we are rapidly moving from an 80% Assisted workflow to an 80% Agentic workflow. Are you ready for the transition? We spend 8 weeks helping executives build Agentic product orgs in our next Leadership Cohort. 👉 Apply here: https://lnkd.in/e4qKrm_E Follow ENDGAME to learn more.

The 60-second cheat sheet for explaining AI to your Board. Stop trying to explain Python libraries and parameters to non-technical leadership. They don't care how the engine works. They just want to know how fast the car goes. If you want to get your budget approved in 2026, explain AI based on what it delivers. Save this 5-step roadmap for your next strategy meeting: 🟢 AI & ML: Turns your data into decisions. 🟢 Neural Networks: Handles complex business pattern detection. 🟢 Gen AI: Generates content and code at scale. 🟢 AI Agents: Executes complex tasks autonomously using tools and function calling. 🟢 Agentic AI: Automates entire processes with AI through long-term autonomy and multi-agent collaboration. Right now, most of your competitors are stuck at the Gen AI layer. They are just using it to write faster. The companies that will dominate the next decade are pushing into the Agentic AI layer. They aren't just optimizing tasks; they are delegating entire workflows. Are you building a strategy for the chatbox, or for the workforce? See comments for the cheat sheet. ________________________________________________ Follow ENDGAME to accelerate how to use AI in your organisations. Thanks to Alex Barády for the inspiration for this post. Follow Jonny Tooze for more insights on AI.

If everyone has access to the exact same AI, how do you build an unfair advantage? That is the question every executive needs to answer this year. When leaders look at this whiteboard, they usually get overwhelmed by the terminology. But you don't need to be a developer to read this chart. You just need to look at it as a "Buy vs. Build" matrix. Here is how to allocate your focus: The Interface (The Inner Boxes): ChatGPT is an amazing tool. But if it is the center of your strategy, you don't have an advantage. You have a $20/month subscription that your competitors also have. You are just renting productivity. The Engine (The Middle Boxes): Generative AI and LLMs are where the real enterprise value is created. The smartest companies are operating here. They take these powerful models, connect them to their proprietary data, and build custom workflows that are impossible for competitors to replicate. The Science (The Outer Boxes): Machine Learning and Neural Networks are foundational. But unless you are a tech giant, you don't need to build from scratch here. You leverage the science to power the engine. The companies winning in 2026 aren't competing on who can prompt ChatGPT the best. They are competing in the middle zone. Where is your team spending their time? __________________________________________________________________ Follow ENDGAME if you care about faster delivery, higher adoption, and measurable returns from AI – not dashboards full of vanity metrics. Thanks to Alex Barády for the spark behind this one. Follow Jonny Tooze for more straight-talk on AI, product, and delivery.

The blueprint for achieving 3x development speed in 2026. We get 2-3 calls a week from executives who are completely stuck. They bought the AI tools, trained the staff, and applied AI to every step of their dev cycle. But they haven't seen any real speed gains. The Reality Check: Optimizing your legacy PDLC doesn't work. To get the 2-3x speed investors are demanding, you have to shift to Agentic Development. Save this blueprint to align your leadership team. How the paradigm shifts: 1. The AI Delivers: Agents handle the program plans, tech specs, raw code, and test cases. 2. The Humans Decide: Your team shifts to high-level governance, dictating scope, standards, quality, and risk. When you combine Agentic execution with automated handovers, you see the massive productivity gains shown at the bottom of the board (up to 70% in QA). Agentic coding works for individuals today. The companies that figure out how to make it work for teams will win tomorrow. Want to learn how to scale Agentic AI? Join our ENDGAME AI-native leadership cohort: 👉 Apply here: https://lnkd.in/e4qKrm_E Follow Jonny Tooze for more straight-talk on AI, product, and delivery.

Everyone is debating whether AI will take jobs. The CEO of Anthropic just put a number on it: 50% in just 5 years. Dario Amodei is not some LinkedIn commentator farming engagement. He builds the technology. And he said: "We have a duty and an obligation to be honest about what is coming." Half of entry-level white-collar roles. Eliminated. Not in a decade. Within 1 to 5 years. You can panic about that, or you can do something about it. I have spent the last 12 months building AI-native systems every single day. Writing production code. Shipping products. Training engineering leaders. Here is what I know for certain: the people who learn AI skills now will not be replaced by AI. They will be the ones replacing the old way of working. So I put together a list of the 10 best free AI courses available right now. Every major provider - Anthropic, Google, Meta, NVIDIA, Microsoft, OpenAI, IBM, AWS, DeepLearning.AI, Hugging Face. All free. No excuses. Start there! And if you lead an engineering team and want to go deeper - not just learn the tools but redesign how your entire organisation builds, ships, and operates - that is exactly what we do in the ENDGAME AI-Native Engineering Leadership Cohort. 👉 8 weeks. 👉 2x faster time-to-market. 👉 40-60% productivity lift. 👉 Cohort 2 starts 20 April. The shift is here. The question is whether you are skilling up or standing still. 👉 Free courses are in the image below. Save it. Share it. Send it to your team. ♻️ Follow Jonny Tooze for more content like this, and repost this if someone in your network needs to see this before it is too late. Apply for ENDGAME's April Cohort here: https://lnkd.in/evkJ4PE9

AI didn't give your team more free time. It just gave them a new way to burn out. The most dangerous myth in business right now is that "AI productivity" creates infinite capacity. Leaders are looking at dashboards showing massive output, but they are ignoring the human cost behind the screen. Your team isn't working less; they are just working differently...and it is draining them. If you aren't actively managing your AI workflows, your team is falling into these 3 traps: 1. Expansion: Doing the jobs of three different departments because the AI lowers the barrier to entry. 2. Ambient Work: Constantly checking AI outputs after hours. It feels like "just monitoring," but it prevents the brain from ever actually logging off. 3. Attention Overload: Fielding pings from 15 different automated workflows destroys the ability to focus on a single strategic problem. Productivity is not about doing everything faster. It is about doing the right things with focus. Leaders must step in. Enforce Sequencing. Batch the agent notifications. Protect your team's deep-focus windows. AI scales output. It does not scale human endurance. Lead accordingly. ___________________________________________________________ Follow ENDGAME if you care about faster delivery, higher adoption, and measurable returns from AI – not dashboards full of vanity metrics. Thanks to Alex Barády for the spark behind this one. Follow Jonny Tooze for more straight-talk on AI, product, and delivery.

Stop calling everything your team does with AI "AI coding." There are three distinct levels. Most organisations are stuck on the first one and do not realise the gap. 1️⃣ Level 1: Vibe Coding. You describe what you want in plain English. The AI generates code. You test and iterate. 🛠️ Tools: Bolt, Lovable, Replit, V0. ⚡Speed: 10x faster prototyping. 🎯 Reality: Speed over precision. Great for validating ideas. Not production-grade. 2️⃣ Level 2: AI-Assisted. Your developers write code with live AI support. Intelligent completions, suggestions, debugging. 🛠️ Tools: Cursor, GitHub Copilot, Kiro, Continue. ⚡ Speed: 25-30% productivity increase. 🎯 Reality: This is where 90% of engineering teams are right now. It is better. It is not transformational. 3️⃣ Level 3: Agentic. AI agents complete entire development tasks autonomously. They plan, write, test, and iterate. 🛠️ Tools: Claude Code, OpenAI Codex, Gemini CLI. ⚡ Speed: 250% faster for complex development work. 🎯 Reality: This is where the step change happens. Not incrementally faster. Fundamentally different. The gap between Level 2 and Level 3 is not a tool upgrade. It is a delivery model redesign. Your engineers cannot get there by installing a new plugin. They get there by rethinking how work flows through the entire product development lifecycle. Which level is your engineering team operating at today? 1, 2, or 3? 👉 Follow Jonny Tooze for the straight talk on AI-native engineering. ♻️ Repost if your team needs to see the difference between assisted and agentic.

You cannot buy your way to a 2x development speed. 90% of companies that are failing at adoption now treat AI development like a standard IT rollout. They buy the licenses, force the adoption, and wait for the company to magically move faster. That is a dangerous illusion. Buying a tool does not change your delivery speed. The winners of 2026 aren't just giving developers a chatbot; they are redesigning the entire workflow to leverage AI for the hardest parts of the job. What real AI development looks like: - Agentic coding for refactoring at scale. - Enabling super rapid MVP prototyping. - Using TDD and AI to ship with full confidence. If your AI strategy fits in the top chart, you are just optimizing a legacy process. If it fits in the bottom chart, you are actually transforming your business. Want to learn how to scale Agentic workflows safely? Join our next ENDGAME Leadership Cohort. We give you the exact blueprints to transform your product org. 👉 Apply here: https://lnkd.in/e4qKrm_E Follow Jonny Tooze for more straight-talk on AI, product, and delivery.

When most leaders start their AI journey, they ask: "What problem does this solve?" It is a completely normal place to start. But it can also keep your team stuck in the weeds. Solving a single problem is a tactic. Building an entirely new organizational capability is a strategy. If you want to build a true competitive moat this year, try shifting the question to: "What capabilities do we need to build to outcompete the market?" To do that safely, you don't just buy a software subscription. You build an organizational engine. Here are the 4 layers every AI-native company is building right now (starting from the bottom up): 1. Data & Technology Foundation The Goal: Connect AI to your existing business. The Focus: Getting your Data Infrastructure and APIs & Integrations ready so models can actually read your company's data safely. 2. AI-Assisted Development The Goal: Ship products at competitive speeds. The Focus: Using Agentic Coding and RAG Implementation to give your development teams a massive speed boost. 3. Governance & Operating Model The Goal: Scale your automation safely, without breaking things. The Focus: Protecting the business with Evaluation Frameworks, Cost Management, and clear Guardrails. 4. AI-Powered Experiences The Goal: Change how employees and customers interact with you. The Focus: This is the fun part, deploying Automated Workflows and autonomous AI Agents. It is completely okay if your company is only on Layer 1 right now. The most important thing is knowing there is a clear roadmap to the top. Save this blueprint to align your leadership team on the bigger picture. _____________________________________________________________ Follow ENDGAME if you care about faster delivery, higher adoption, and measurable returns from AI – not dashboards full of vanity metrics. Thanks to Alex Barády for the spark behind this one. Follow Jonny Tooze for more straight-talk on AI, product, and delivery.

Stop looking at Claude as a software subscription. Look at it as your new org chart. Most companies are only using 33% of Claude's capability. They treat it like a chatbot. But if you look at Anthropic’s new product lineup, "Chatting" is just the entry point. They haven't just released features; they’ve effectively built a 3-tier workforce. If you want to be an AI-Native organization in 2026, you need to deploy all three: 1. Claude (The Thinking Partner) The Function: "Answer any question". The Role: Your Consultant. You ask a question, and it answers like a very smart colleague. Perfect for strategy, board prep, and summarizing contracts. 2. Claude Code (The Engineering Partner) The Function: "Build it for me". The Role: Your Senior Developer. It doesn't just chat; it reads your entire software codebase. It runs the Plan ➝ Code ➝ Test ➝ Fix loop autonomously until the finished product is delivered. 3. Cowork (The Operations Partner) The Function: "Handle any work". The Role: Your PM or Ops Lead. It brings that exact same self-correcting loop to knowledge work. It uses plugins to execute tasks across sales, finance, legal, and marketing. Claude helps you think. Claude Code helps you build. Cowork helps you operate. If your team is only using the first column, you are leaving the real leverage on the table. Want to go deeper? We are helping leaders implement this exact stack in our next ENDGAME AI-Native Leadership Cohort. 👉 Apply here: https://lnkd.in/e4qKrm_E Follow Jonny Tooze for more straight-talk on AI, product, and delivery.

The 6-step blueprint for Agentic Engineering. 90% of companies measure AI adoption. The 10% that actually win measure outcomes. We just used this exact framework to take a client from a 2-year backlog to a fully delivered roadmap in 4 months. Save this roadmap to align your leadership team: 1. Pick the right bets: Start with test generation and bug triage. 2. Prove it in production: No sandboxes. Real projects. Real deadlines. Measure actual cycle time. 3. Codify everything: Agents only follow what is written down. Standardize your specs and rulebooks. 4. Build the shared layer: Invest in shared toolchains and security guardrails first. 5. Evolve the roles: Move your engineers from writing code to specifying intent. They are architects now. 6. Structure the team: Agents do the delivery work. Humans set direction, review, and decide. Stop counting software seats. Start measuring delivery speed. Need the architectural support to implement this Dozens of CTOs are currently learning how to scale Agentic workflows in our AI-Native Leadership Cohort. 👉 Apply here: https://lnkd.in/e4qKrm_E ♻️ Repost to get your leadership team speaking the same language.  Follow Jonny Tooze for more straight talk on AI-native delivery.

Investor patience for AI experimentation just ran out. I am hearing the same conversation in every boardroom. "We gave you the budget. We gave you the tools. We gave you 18 months. Where is the delivery speed?" The answer, in almost every case, is silence. 📊 McKinsey's data has not changed. 88% of companies bought AI tools. 6% actually changed how they work. But the pressure has changed. Dramatically. In February, investors were asking questions. In March, they are making decisions. ⚡ Here is what shifted: 🔹 AI-native competitors are shipping products in 6 weeks that used to take 6 months 🔹 The cohort of companies that redesigned their delivery models are now showing 2-2.5x speed improvements with hard data 🔹 Boards can see the gap between what they funded and what they got The companies still running the same agile rituals, the same spec-to-ticket pipelines, the same manual QA processes with an AI assistant bolted on are not transforming. They are decorating. And investors can now see the difference. The 6% that moved did three things: 1️⃣ They stopped treating AI as a developer tool and started treating it as a delivery model 2️⃣ They replaced their pipeline, not just their toolchain 3️⃣ They measured time-to-market, not adoption rates The gap between the 6% and the 88% is not closing. It is accelerating. Every quarter that passes, the companies that redesigned pull further ahead. If your board is still funding AI experiments without a delivery transformation plan, that conversation is coming. And it will not be a question next time. 👉 Follow Jonny Tooze for the straight talk on AI-native delivery. ♻️ Repost if your CTO needs to see this before the next board meeting.