Future Of Work

Marketing careers aren’t dying because of AI. They’re splitting because of it. I was having a conversation yesterday about what happens to the generation of marketers who built their careers through digital, transformation and the convergence of marketing and technology. I think we are entering a period where that broad digital marketing profile will increasingly split into two very distinct career paths. For the last decade or so, digital marketing rewarded people who could operate across disciplines. Digital was still evolving, so companies needed people who could sit in the middle and make sense of it. That middle is becoming harder to define. 1. Marketing Systems Leaders: One path is moving deeper into data, technology and product. These marketers will increasingly need to understand marketing technology, customer data, AI, automation, product experience and the systems that sit behind acquisition, engagement and retention. Product marketing increasingly sits naturally in this space. 2. Marketing Craft Leaders: The other path is moving deeper into marketing itself:brand, positioning, consumer understanding, communications, creative strategy and the ability to build distinctiveness in a market where execution is becoming increasingly commoditised. I'm seeing that companies are becoming much more deliberate about what's the role of marketing and the marketing hires. AI is accelerating this shift. There will still be exceptional people who can bridge both worlds. In fact, I suspect they will be disproportionately valuable. But the generic middle is going to become much harder to defend. Where do you think the next generation of digital and transformation marketers will build their depth?

IKEA automated the task, not the people. I have always believed that the purpose of automation should be to remove repetitive work, but IKEA’s approach shows what that principle looks like when a company applies it at scale. Its AI chatbot, Billie, began handling routine customer enquiries, which could easily have turned into a familiar cost-cutting exercise. Instead, IKEA reskilled 8,500 call-centre employees in areas such as remote interior design, digital sales, relationship-building, and more complex customer support. What I find especially interesting is that IKEA did not treat those employees as redundant simply because part of their work had become automatable. It looked at the knowledge they had already built about customers and products, then asked how that experience could create more value elsewhere. The wider remote sales channel generated €1.3 billion in revenue in 2022. We cannot attribute that entire figure to reskilling alone, but the strategic direction still matters. In my Human+AI Equation, technology provides scale and speed, while people provide judgment, creativity, and connection. IKEA did not choose between the two. It redesigned the work around both. The best automation strategy does not begin by asking, “How many people can we remove?” It begins by asking, “What more valuable work can our people do now?” How would your AI strategy change if every automation plan also required a reskilling plan? #HumanAgentOrchestrator #AITransformation #HybridManagement #WorkforceReskilling

🚨 Zero Trust for AI Agents Anthropic just released "Zero Trust for AI Agents." As we're thinking about agentic permissions, applying a Zero Trust discipline is critical to secure adoption. AI agents interpret goals, call tools, chain actions, delegate to other agents, and maintain context across sessions. The trust surface is different. The paper introduces "least agency" — a concept that OWASP has been promoting — and the distinction from least privilege is worth sitting with. 👉 "Least privilege" asks what an identity can access. 👉 "Least agency" asks what an agent can do, under what conditions, with which tools, and with what level of oversight. Autonomous agents introduce action risk alongside access risk — and the boundaries around behavior need to be architecturally enforced, not assumed. The paper includes a design test worth writing down: 🔥 Does the control make the attack impossible, or merely tedious?🔥 The practical controls follow directly from Zero Trust fundamentals — cryptographic agent identity, short-lived credentials, tool allow-listing, sandboxed execution, and full traceability from prompt to action to outcome. None of this is new doctrine. It's existing architecture applied to a harder problem. Full disclosure: the paper cites NIST SP 800-207 on Zero Trust Architecture and the CISA Zero Trust Maturity Model, both of which I co-authored during my time supporting Federal Zero Trust efforts at CISA. Zero Trust is built for a world where we have to remove implicit trust. Agentic AI is the next version of that same problem — valid identities, valid credentials, legitimate-looking actions, and still no basis for assumed trust. Access is earned. Actions are constrained. Agency must be governed. 👉🏼 Link to Anthropic's paper in the comments.

Confusing "product owner" with "product manager" is like confusing "Scrum Master" with "engineering leader." Product Owner is a role in Scrum. Product Manager is a career. This distinction matters more than most companies realize. When you only recognize Product Owner roles, you're signaling that product work is just about managing backlogs and running ceremonies. That's tactical execution, not strategic leadership. The actual career path looks like this: Associate PM → PM → Senior PM → Director → VP → CPO. And as you progress, your work fundamentally shifts. Early in your career, you might spend 70% of your time on tactical work, like writing requirements, managing stakeholders, coordinating releases. But as you advance, that flips. Senior product leaders spend 70% of their time on strategy, vision, market positioning, and organizational alignment. Companies that don't create proper career paths for product managers struggle to attract and retain top talent. Why would an experienced PM take a "Product Owner" role when they're ready to drive strategy at the director level? If you want to build a strong product organization, start by recognizing product management as the strategic function it is, not just a project coordination role. On podcast episode 252 of the Product Thinking with Melissa Perri, I explore the difference between Project and Product management, which have also created many misconceptions in the industry. Attached you can see the visual from my book “Escaping the Build Trap” from 2018. How well does this model still hold up in 2026?

AI security/securing the use of AI is going to kill me. I use Claude Code almost daily. It's a problem.... Here's what I have to change AGAIN this week. Security researcher Ari Marzuk disclosed 30+ vulnerabilities across AI coding tools. Cursor. GitHub Copilot. Windsurf. Claude Code. All of them. He called it IDEsaster. The attack chain includes prompt injection, hijacking LLM context, and auto-approved tool calls executing without permission. Then, legitimate IDE features are weaponized for data exfiltration and RCE. Your .env files. Your API keys. Your source code. Accessible through features you thought were safe. Most studies I read claim that around 85% of developers now use AI coding tools daily. Most have no idea their IDE treats its own features as inherently trusted. 𝗦𝗼... 𝗮𝗳𝘁𝗲𝗿 𝗿𝗲𝘃𝗶𝗲𝘄𝗶𝗻𝗴 𝗔𝗿𝗶'𝘀 𝗿𝗲𝘀𝗲𝗮𝗿𝗰𝗵, 𝗵𝗲𝗿𝗲'𝘀 𝗜 𝘄𝗶𝗹𝗹 𝗯𝗲 𝗱𝗼𝗶𝗻𝗴... Be warned: All this is SO much easier said than done! Audit every MCP server connection. Checked for tool poisoning vectors where legitimate tools might parse attacker-controlled input from GitHub PRs or web content. Removed servers I couldn't verify. Disabled auto-approve for file writes. The attack chains weaponize configuration files and project instructions like .claude/settings.json and CLAUDE.md. One malicious write to these files can alter agent behavior or achieve code execution without additional user interaction. Move all credentials to a secrets manager. No .gitignored .env files in agent-accessible directories. API keys live in 1Password CLI. Environment variables inject at runtime through a wrapper script the LLM never sees. Start running Claude Code in isolated containers. Mounted volumes limited to specific project directories. No access to ~/.ssh, ~/.aws, or ~/.config. If the agent gets compromised, blast radius stays contained. Enable all security warnings. Claude Code added explicit warnings for JSON schema exfiltration and settings file modifications. These exist because Anthropic knows the attack surface. Add pre-commit hooks for hidden characters. Prompt injections hide in pasted URLs, READMEs, and file names using invisible Unicode. Flag non-ASCII characters in any file the agent might ingest. The fix isn't to stop using AI coding tools. The fix is to stop trusting them implicitly. What controls do you have for AI tools with write access to your codebase? 👉 Follow for more AI and cybersecurity insights with the occasional rant #AISecurity #DevSecOps

At Meta, there's a famous poster of a rocking horse that says "Do not mistake motion for progress." Recently I saw an AI study proving exactly why: teams using AI tools felt 20% more productive while actually being 19% less productive. How? They spent more time prompting, waiting, and reviewing AI output. Less than 44% of AI suggestions were accepted without modification. Yet they felt 20% faster while going backwards. The scariest part: without measurement, these teams would have doubled down. They felt productive. Their managers saw more output. Everyone was happy except the business metrics. Here's how to avoid becoming another AI casualty in 2025: 1. Set one primary metric per team per quarter. Just one. Not ten KPIs. Not a balanced scorecard. One number that moves the business. Activation rate. Retention. Gross margin. Pick one. While everyone's deploying AI tools based on how they feel, companies that built billion-dollar empires measure everything. Procter & Gamble (the $400B company behind Tide, Gillette, and 100+ other brands) runs every initiative against one primary metric. A new sales process? Conversion rate. Marketing campaign? Revenue attributed. Clear pass/fail criteria. The lesson: features are motion. Metric improvement is progress. 2. Run time-boxed trials with control groups for every AI tool. Fortune 500 companies force teams to define success criteria upfront. Before you build, you write. Before you deploy, you measure. UPS learned this with their route optimization. They measured miles per route. Ran pilots site-by-site. Scaled only after cutting 6-8 miles per route. Now saves 100 million miles annually. 3. Cap work-in-progress to force actual completion. AI makes it trivially easy to start new things. Generate a proposal. Draft ten email campaigns. Create fifteen dashboard variations. More motion, everywhere. But starting isn't finishing. Toyota learned this decades ago: limit work-in-progress. Cap initiatives per team. You can't start something new until you ship or kill something old. Why? Because ten half-built features are worth less than one that actually ships. AI amplifies this problem - it's never been easier to create motion that looks like progress but delivers nothing. — 2025 is the year of AI-accelerated motion. Notice the pattern: every successful company makes it HARDER to ship, not easier. They add friction. They demand evidence. They stop more than they start. Because they learned what that rocking horse teaches: motion without progress is just expensive theater. While everyone else rides the AI rocking horse, you'll be the one actually moving forward.

Monthly Book Review: Two books for building AI agents - from very different angles. 📘 𝐁𝐮𝐢𝐥𝐝𝐢𝐧𝐠 𝐀𝐠𝐞𝐧𝐭𝐢𝐜 𝐀𝐈 𝐒𝐲𝐬𝐭𝐞𝐦𝐬 (overview, good to start with) This one takes a systems-level view - why agentic systems matter and how they’re shaping the future of autonomy and intelligent applications. In short, it focuses more on big-picture and architecture. It’s more like a guide for people thinking about what agentic AI should become. What stood out to me: - It frames agents not as “chatbots with tools” but as infrastructure for enterprise autonomy - Strong focus on ethics, trust, transparency, and explainability. These are treated as design principles, not just add-ons - Talks about personalization, context-awareness, and decision-making as key challenges - Covers multi-agent collaboration and how agents can plan, adapt, and learn from feedback If you care about the long-term view, this one is worth digging into. 📙 𝐁𝐮𝐢𝐥𝐝𝐢𝐧𝐠 𝐀𝐈 𝐀𝐠𝐞𝐧𝐭𝐬 𝐰𝐢𝐭𝐡 𝐋𝐋𝐌𝐬, 𝐑𝐀𝐆, 𝐚𝐧𝐝 𝐊𝐧𝐨𝐰𝐥𝐞𝐝𝐠𝐞 𝐆𝐫𝐚𝐩𝐡𝐬 (practical and more advanced) This one is for builders. Especially if you're working on agents that need reasoning, memory, and retrieval grounding. It’s more technical and focused on real implementation. The book walks through how to combine LLMs with: ➤ 𝐑𝐀𝐆 𝐩𝐢𝐩𝐞𝐥𝐢𝐧𝐞𝐬 for document-grounded reasoning ➤ 𝐊𝐧𝐨𝐰𝐥𝐞𝐝𝐠𝐞 𝐠𝐫𝐚𝐩𝐡𝐬 for structured, symbolic understanding ➤ 𝐌𝐮𝐥𝐭𝐢-𝐡𝐨𝐩 𝐩𝐥𝐚𝐧𝐧𝐢𝐧𝐠, 𝐭𝐨𝐨𝐥 𝐮𝐬𝐞, 𝐚𝐧𝐝 𝐬𝐜𝐡𝐞𝐦𝐚 𝐚𝐥𝐢𝐠𝐧𝐦𝐞𝐧𝐭 - the essential components for smart workflows What I found helpful: - Very hands-on, with code, architecture diagrams, and real examples - Focused on making agents reliable and grounded in actual knowledge, not just generating text - Covers planning logic, memory structures, and how to build better reasoning flows It’s more advanced than the first book, but still practical and useful for real-world builds. 𝐌𝐲 𝐫𝐞𝐜𝐨𝐦𝐦𝐞𝐧𝐝𝐚𝐭𝐢𝐨𝐧: ➤ Start with the first if you want to understand the broader vision and systems thinking. ➤ Pick up the second if you're focused on reasoning workflows and building better LLM-powered agents. They go well together. Two perspectives that cover both what to build and how to build it. Let me know if you’ve read either. Curious what others are exploring in this space. Links to both books below: ✔️ Building Agentic AI Systems by Anjanava Biswas and Wrick Talukdar  https://packt.link/HiXtG ✔️ Building AI Agents with LLMs, RAG, and Knowledge Graphs by Salvatore Raieli and Gabriele Iuculano https://packt.link/KgUPm __________ For more on AI and learning materials, plz check my previous posts. I share my journey here. Join me and let's grow together. Alex Wang #agenticai #aiagents #llms #rag #generativeai

I taught myself machine learning > 10 years ago. If I had to start again today, I wouldn’t touch models, LLMs, or agents first, as many AI experts suggest. I'd start with the math and the code. Ugly truth: 90% of people skip the foundations, then wonder why everything feels like magic or falls apart in production. If you want to be different, actually understand ML, not just copy-paste, this is the roadmap I'd follow: Start with fundamentals: Because no matter how fast LLMs or GenAI evolve, your math, code, and logic will keep you relevant. Here's what you should focus on: 📐 1. Linear Algebra Learn these core ideas: Vectors, matrices, tensors Matrix multiplication (dot products, broadcasting) Transpose, inverse, rank, determinants Eigenvalues & eigenvectors (especially for PCA & embeddings) Projections and orthogonality ✅ Use NumPy to implement everything yourself → Practice matrix ops, dot products, and visualizing transformations with Matplotlib 🔁 2. Calculus Focus on: Derivatives & partial derivatives Chain rule (for backpropagation in neural nets) Gradient descent Convex functions, minima/maxima ✅ Use SymPy or JAX to visualize and compute derivatives → Plot functions and their gradients to develop deep intuition 🎲 3. Probability You need a solid grip on: Random variables (discrete & continuous) Conditional probability & Bayes' rule Joint & marginal probability The Chain rule Expectation, variance, entropy Common distributions: Bernoulli, Binomial, Gaussian, Poisson Central limit theorem The law of large numbers ✅ Simulate simple probability experiments in Python with NumPy → E.g. simulate sampling from distributions 📊 4. Statistics These are must-know topics: Descriptive stats: mean, median, mode, standard deviation Hypothesis testing: p-values, confidence intervals, t-tests Correlation vs. causation Sampling, bias, and variance Overfitting/underfitting A/B testing basics ✅ Use Pandas & SciPy to explore real datasets → Calculate descriptive stats, create histograms/box plots, run t-tests 🔧 Essential Python libraries to learn early NumPy – for vectorized math and fast array ops Pandas – for loading, cleaning, and analyzing tabular data Matplotlib / Seaborn – for plotting and visualizing distributions, relationships, and trends SymPy – for symbolic math and calculus SciPy – for stats, optimization, and numerical methods Use Jupyter Notebooks(to combine math, code, & visuals in one place) 📚 Best resources to nail the fundamentals: ✅ Machine Learning Foundations Math series (ML Foundations: Linear Algebra, Calculus, Probability, and Statistics)-series of 4 courses that I've created together with LinkedIn learning ✅ Hands-On ML with TensorFlow & Keras book by Aurélien Géron ✅ The Hundred-page Machine Learning Book by Andriy Burkov If you want to become an actual ML engineer, not just someone who watches and copies demos, start here. ♻️ Repost to help others💚

Everyone’s suddenly talking about 𝗖𝗼𝗻𝘁𝗲𝘅𝘁 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴. Here’s why it matters. In the AI gold rush, most people focus on the LLMs. But in reality, context is the product. Context engineering is the emerging discipline of designing, assembling, and optimizing what you feed a LLM. It’s the art and science behind how RAG, agents, copilots, and AI apps actually deliver business value. It includes: - What information to surface (data selection, chunking, and formatting) - How to frame the user intent (prompt design, agent memory, instructions) - How to dynamically adapt to each interaction (tool use, grounding, policies) Think of it as the new software architecture but for AI reasoning. And just like traditional engineering disciplines, it’s becoming repeatable, measurable, and mission-critical. 💡The future isn’t just “prompt engineering.” It’s context engineering at scale; where the AI is only as good as the ecosystem of inputs it’s wired into.

For decades, career growth followed a familiar formula: More headcount. More budget. More scope.  That model is changing. In the AI era, careers won’t be built on span of control, they’ll be built on innovation density. Today, anyone - from ICs to execs - can scale their impact without more headcount, more budget, or more time. The playing field is flatter. The differentiator? How fast you can learn, apply, and compound innovation with AI. If you’re thinking about career growth, stop asking: “How can I get more?” Start asking: “How can I innovate more with AI?” The people who rise fast will: See problems through an AI-first lens. Move from manual to scalable. Iterate faster than the rest. Your team size won’t define your trajectory. Your creativity will. Your budget won’t signal your value. Your innovation density will.