What You'll Learn
The New Bottleneck in AI Development
When technical barriers to building software collapse, a paradox emerges: companies gain the power to create anything but often lack the judgment to decide what they should create. This tension formed the core of a leadership panel at Create With conference, where four experts with diverse backgrounds tackled the strategic challenges of AI adoption.
The panel brought together JJ Englert from Tenex focusing on AI governance, Isabelle Amazon-Brown working with international development organizations, Oana May who has coached over 1,000 product and engineering leaders since leaving Meta, and Max Haining whose company 100 School helps organizations convert AI access into genuine capability.
The 'Should We' Question Replaces 'Can We'
Isabelle Amazon-Brown reframed the central premise immediately: "The problem is the word 'can' and I would actually replace it with the word 'should'." Her work in the humanitarian sector illuminates this tension sharply. Organizations serving vulnerable populations face pressure to maximize efficiency and stretch limited funding further, yet operate in contexts where the stakes of getting it wrong are dramatically higher than in commercial settings.
The data Oana May presented reinforces this concern. Research from Pendo in 2019 showed that 80% of features built were rarely or never used. With AI democratizing the ability to build, that ratio likely worsens. Recent engineering research reveals that while AI adoption drives massive increases in code production, actual bottom-line business gains remain minimal - not because teams aren't building more, but because they aren't building the right things.
Where Non-Technical Builders Should Start
Max Haining's advice for non-technical people entering the building space strips away the intimidation factor. The word 'build' itself carries different meanings across audiences - from vibe coding that feels daunting to beginners, to simple task automation for those just getting started with AI.
His recommended first step: map out tasks you'll execute today, identify the most annoying one to do without AI, then use tools like Claude to work through building a solution. The scale doesn't matter initially - it might produce simple copy or evolve into a complete system. The goal is practical engagement that builds confidence and capability incrementally.
Developing Product Judgment at Speed
Product sense - the intuition for what's worth building - traditionally develops through years of experience making product decisions and learning from failures. Oana May has systematized this learning into a framework that accelerates judgment development for new builders.
Her approach centers on asking specific questions before writing any code: Who exactly experiences this problem? How important is it to them? Would they wake up in the middle of the night willing to pay significantly to solve it? Is the addressable market large enough? Most new builders, she observes, start too broad without sufficient specificity about the actual problem, the people affected, or the business viability.
The framework forces precision: understanding not just what could be built, but whether it should be built based on genuine user pain, market size, competitive positioning, and strategic fit. Without this disciplined thinking, democratized building tools simply enable faster creation of products nobody wants.
Research Must Precede Production Building
Isabelle Amazon-Brown emphasized that research and validation should precede building in production contexts, though she distinguished this from prototyping for testing and alignment. The enthusiasm of non-traditional builders entering the space is valuable, but fundamental product principles developed over decades still apply - arguably with even higher stakes now.
In her sector, inadequate research before building can have serious consequences for already vulnerable populations. But even in commercial contexts, building without research wastes resources and creates technical debt. The speed at which AI enables creation makes the discipline of research and validation more critical, not less.
The Strategic Value of Prototyping
While production building requires research and validation, the panel recognized distinct value in prototyping even when code won't ship. Prototypes improve organizational alignment, test assumptions quickly, and enable stakeholders to respond to tangible artifacts rather than abstract specifications.
This distinction matters for governance frameworks. Companies shouldn't gate all building behind extensive approval processes, but should differentiate between experimental prototyping that builds capability and organizational learning versus production systems that serve customers and carry operational risk.
Converting Access into Capability
Max Haining's work with 100 School focuses specifically on the gap between having AI tools available and actually using them effectively to change how work gets done. Their flagship program, 15 Days of AI, upskills entire companies to raise baseline proficiency, then helps organizations move beyond isolated usage toward genuine transformation in working methods.
This progression - from access to usage to capability to changed work patterns - provides a framework for measuring AI adoption maturity. Many organizations stall at the 'access' or 'usage' stages without building systematic capability or fundamentally rethinking workflows.
Leadership Modeling and Safe Experimentation
The panel consensus held that leadership behavior shapes AI adoption more than tool availability. Leaders must model experimentation, continuous learning, and intelligent risk-taking rather than simply encouraging it in others.
Creating psychologically safe environments for experimentation within companies requires clear boundaries around acceptable risk, processes for learning from failures, and cultural norms that reward thoughtful experimentation even when initial attempts don't succeed. Without this safety, teams default to cautious behavior that limits learning and capability development.
Governance Without Stifling Innovation
Balancing governance needs with innovation speed presents a persistent challenge. Organizations need frameworks that enable safe experimentation while preventing serious risks - regulatory violations, security breaches, reputational damage, or harm to users.
Effective governance focuses on risk levels and impact rather than blanket restrictions. Low-risk internal experiments might require minimal oversight, while customer-facing applications need robust validation. Clear criteria for risk assessment, appropriate approval processes for different risk levels, and continuous learning loops create governance that enables rather than blocks innovation.
Building a Learning Organization
The shift toward accessible AI building tools fundamentally changes organizational learning requirements. Companies need systematic approaches to developing judgment, product sense, and strategic thinking across broader populations.
This includes formal training programs like those Max Haining runs, but also peer learning, communities of practice, documentation of decisions and outcomes, and explicit frameworks like Oana May's system for evaluating what to build. The goal extends beyond individual capability to organizational capability - shared language, common frameworks, and collective judgment that improves over time.
Practical Next Steps for Leaders
For leaders navigating this transition, the panel's guidance coalesces around several themes: start with clear use cases that address real pain points, invest in developing judgment alongside technical capability, create safe spaces for experimentation with appropriate boundaries, model continuous learning yourself, and remember that building faster only creates value when directed by strategic thinking about what should be built.
The democratization of building tools represents genuine opportunity, but only for organizations that develop the complementary capabilities in research, validation, strategic thinking, and organizational learning. The companies that thrive won't be those that build the most - they'll be those that combine newfound building capability with discipline about building the right things.
