BlueSkyWork Blue-Sky Thinking. Building What's Next

Blue Sky Insights

Ideas about building, leading and redesigning product organizations in the age of AI

Over the past few years, I've been exploring how AI changes the way we build products, organize Product & Engineering teams, and ultimately how we lead software companies.

These ideas combine lessons from years of building and scaling Product & Engineering organizations with hands-on experiments building AI-native products and new ways of working. Fieldwork collects what I've built and what each piece taught me.

I share these ideas as LinkedIn posts, where they benefit from discussion and challenge. This page brings them together in one place - from first principles and operating models to practical field notes and leadership lessons.

First Principles

Foundational articles exploring why AI changes the economics and assumptions behind modern product organizations.

  • The Bottleneck Didn't Disappear. It Moved.

    Building got faster - but that only exposes how much still depends on product's own capacity and the clarity of the bets it's building toward.

  • AI Agents May Accelerate More Than AI Agents

    AI Agents may have an important second-order effect: accelerating democratized workflow automation by getting organizations to rethink work that has remained unchanged for years.

  • From Personal AI to Organizational AI

    The bigger AI opportunity may be moving from personal AI that improves individual productivity to structural AI that redesigns how teams work together.

  • AI may be accelerating confidence faster than competence

    AI makes building easier than ever, but judgment has not become easier with it. Are we sometimes confusing reduced friction with reduced complexity?

  • Reflections from SaaSiest 2026 Malmø about AI adoption

    At SaaSiest, the AI conversation had shifted from why AI matters to how leading SaaS companies are redesigning their operating models - but the gap is widening.

  • Why SaaS Isn't Going Anywhere

    My take: AI is changing the build-vs-buy equation, but building software isn’t the same as building enduring SaaS. The bar for SaaS isn’t disappearing - it is getting higher.

  • The Difference Between AI Adoption and AI Transformation

    Most organizations are good at evolutionary change. But realizing AI’s bigger potential requires something more revolutionary - redesigning work, not just optimizing it.

  • Be a Scientist, Not a Magician

    Forget the magic prompt and the perfect incantation. Working well with AI is more science than wizardry - experiment, adjust, learn, and build the right system around it.

  • Average AI Adoption May Already Mean Falling Behind

    AI is evolving so fast that the “middle of the pack” may not be a safe place to be. With a technology this transformative, average adoption could quietly become falling behind.

  • The Lean Startup, Reimagined for the AI Era

    AI is changing what “lean” can mean for startups - compressing the cost and speed of experimentation, and challenging old assumptions about how we build, fund and govern innovation.

AI-Native Product Organizations

Articles about operating models, organizational design and how Product & Engineering teams evolve in the age of AI.

Building AI-Native Products

Hands-on lessons from building AI-native products, tooling and engineering practices.

  • Trying to Talk My Own AI Coach Into Leaking Its Secrets

    Why one failed attempt didn’t reassure me, and what it taught me about where the secrets live when AI sits at the core of a product.

  • How We Created Learning Loops Without User Data

    Early-stage AI products rarely have enough user data. The advantage comes from finding representative real-world scenarios and turning them into learning loops before the data arrives.

  • Designing AI Systems in Layers

    Inconsistent LLM outputs are often an architecture problem, not a model problem. Designing instruction and context in layers makes AI features easier to debug, maintain and scale.

  • Measuring the "Feel" of AI

    How do you build an AI that doesn’t just answer correctly, but feels right? A practical look at the two kinds of evaluation I used when testing Work2gether AI coaching models.

  • Why AI Products Benefits From Model-Agnostic Architecture

    Foundational models are evolving too fast to hardwire your product to one of them. A model-agnostic architecture makes it easier to experiment, switch models and keep pace as the landscape changes.

  • Testing LLMs Means Thinking Differently

    LLMs aren’t deterministic software, so testing them needs a different playbook. Think acceptable ranges, golden sets, guardrails and iteration - not exact outputs.

Field Notes

Practical observations from building AI-native products, experimenting with new tools and workflows, and exploring how AI changes the way we create products.

Product Leadership & Operating Models

Timeless product leadership and SaaS operating model lessons.

  • The Real Test for Forward Deployed Engineering

    Part two of the FDE thread: the simple test I would apply before scaling it, and why the test only passes if you build for it.

  • Forward Deployed Engineers: Same Idea, or Something Different?

    A closer look at where FDE actually diverges from the empowered engineering teams I've spent my career building - and what that reveals about ownership.

  • Should Product be involved in go-to-market? - Only if you want to succeed

    The strongest Product, Engineering and GTM organizations don’t just build faster - they operate as one learning system, turning insights into better decisions together.

  • The Three Horizons of Product Investment

    Balancing today’s core, tomorrow’s growth and the longer-term future is one of product leadership’s hardest trade-offs. I find that The Three Horizons model helps make those choices visible.

  • Clarity of Direction Doesn't Make Execution Simple

    Good product principles are often simple. Executing them in a real SaaS business means navigating customers, revenue, incentives and timing - and where the hard work really begins.

  • Scaling SaaS Is Like Flying an Airplane

    Scaling SaaS can feel a lot like flying an airplane: takeoff, turbulence, then hopefully cruising altitude. The trick is knowing which stage you are in - before someone starts serving snacks during takeoff.

  • The Fastest Way to Waste SaaS Runway

    When Product, Engineering and GTM pull in different directions, chaos compounds. Shared strategy, ruthless prioritization and a clear operating rhythm can make growth compound instead.

  • De-risking B2B SaaS Growth

    Scaling B2B SaaS is not just about adding people and moving faster. It is about building the right disciplines to turn fragile growth - bigger teams, roadmaps and burn - into resilient growth.

  • From Founder-Led to Founder-Inspired

    As SaaS companies scale, the founder can shift from greatest asset to biggest bottleneck. The challenge is moving from founder-led to founder-inspired - without losing the vision that got you there.

  • Documentation Is an Accelerator, Not Overhead

    Agile was never an excuse for poor documentation. Done right, documentation helps decisions and knowledge scale - and becomes even more valuable when AI can act on it.

High-Performance Organizations

Broader reflections on leadership, execution and high-performing teams.

  • Do AI Coding Agents Reward Collaborative Teams?

    A reflection inspired by Microsoft's Scott Hanselman's analogy between AI coding agents and open source collaboration - and what it might mean for collaborative product & engineering teams.

  • Build Organizations Like Olympic Rings

    What if resilient organizations should look more like Olympic rings than dominoes? A simple metaphor for why overlapping skills and responsibilities may matter more than ever in the age of AI.

  • Execution Debt: The Hidden Scaling Constraint

    Execution Debt builds quietly as friction accumulates in decisions, ownership and alignment. What looks like a normal scaling problem may actually be that debt starting to accrue interest.

  • The Modern Manager Is Becoming an Integrator

    “Who do I refer to now?” is a common question after a re-org. But the real confusion may not be reporting lines - it is where ownership and decision authority actually live.

  • Focus Is the Missing Half of Empowerment

    Empowered teams don’t just need autonomy and accountability - they need focus. Without clear priorities and trade-offs, ownership can easily turn into diffusion.

  • Empowered teams need both autonomy and accountability

    Giving teams more freedom is only half the equation. Without clear accountability, autonomy can quickly turn into friction, confusion and eroding trust.

  • Your Team Doesn't Need a Hero

    High-performing teams aren’t built by hiring more heroes. They are built by creating the conditions where people can collaborate, perform and succeed together.

  • Great collaboration doesn’t care where you sit

    Great collaboration doesn’t happen automatically when teams become distributed. Trust, engagement and performance across locations need to be designed for - and led differently.