Framework
The 70/30 Software Stack
Seventy percent of every new software application is commodity: authentication, user management, CRUD operations, data pipelines, notifications. It was never innovative. Every dev shop had internal boilerplates for it. The competitive advantage in software has always come from the 30% that's unique. In 2026, this matters more: any decent CRM can be built in a day. Automate the 70%; focus all human effort and genuine IP development on the 30%.
How it works
- 1The original thesis behind Lio (then New Binary's internal tooling): build an abstraction layer over a proven tech stack that generates all commodity components automatically, freeing development for differentiation.
- 2Dev shops had always known this — their boilerplates and module libraries were the manual version of the same insight. The question was why anyone was still writing the login system by hand.
- 3In 2026 the point is sharper: if your moat is building software, you no longer have a moat. A capable developer with AI can produce a CRM in a day. The only defensible edge is intellectual property that can't be replicated by a prompt.
- 4For Canada specifically: the nation invests ~$53B/year in R&D but commercializes a tiny fraction of the resulting IP. As software-as-moat disappears, deep IP becomes the only viable competitive position — making commercialization gaps existentially more important.
- 5For AI tool startups: your 70% (the commodity layer) is at constant risk from model providers. Build your moat in the 30% — workflow depth, proprietary data, accumulated customer context.
Hear it on the podcast
Most AI Prototypes Die Before LaunchListen to the episodeMore from this episode
The Four-Category AI Coding TaxonomyAI coding tools fall into four distinct categories with different users, capabilities, and failure modes: (1) generalist code assistants (Copilot, Cursor) for developers; (2) prototyping tools (Lovable, Bolt, v0) for fast front-end visualization; (3) lifecycle agentic flows (Devin) for engineering teams across the full SDLC; (4) MVP builders (Lio) for full-stack requirements-to-scaffold generation. Knowing which category you're in tells you what to expect, where you'll hit walls, and where to go next.Requirements-First DevelopmentThe CTO's core job is not writing code — it's translating business objectives into technical requirements, then ensuring the code solves the right problem. Start there. Generating code before generating requirements produces something technically functional that solves the wrong problem. The 70% wall is the symptom; deferred requirements are the cause. Requirements elicitation — guided prompting to extract what you actually need — is what separates MVPs from prototypes.
Get the next framework first
One email a week with the frameworks, tools, and founder stories from the podcast. Free.
More frameworks for Building the Product
The Flow-First Product PrincipleDesign your product around protecting momentum, not just delivering features. The most valuable intervention in a creative work session isn't the best feature — it's the one that keeps someone from abandoning their project entirely. On Demand Human is built on this principle: 15 minutes of expert help at the right moment is worth more than any feature upgrade, because it preserves the flow state that makes all other work possible.The Virtuous Expertise LadderIn peer expertise marketplaces, today's beginner becomes tomorrow's best helper. The person who just solved a problem is more empathetic, more contextually accurate, and more effective at explaining the fix than a senior expert who solved it years ago. This creates a self-sustaining supply side: as a community matures, freshly-graduated members flow back into the marketplace as helpers. Design for this natural progression — it's both your supply mechanism and your community glue.Build Backwards from Business RequirementsWoz's core development philosophy: architecture decisions, security guardrails, and compliance requirements are defined at the start from a CTO lens — not added iteratively after prototyping proves the concept.Design Systems ThinkingStop designing individual screens. Start designing systems of experiences. Every interface element is a pattern with explicit rules and constraints — not a one-off creative decision. This shift from instance-level to system-level abstraction is what enables consistent product experiences at scale and makes AI generation enterprise-appropriate.Prototype and PruneAI is a proxy for skill, not taste. Generate multiple enterprise-appropriate prototypes using real design system context, then let a human with taste prune down to the right one. Coined by Jules at Meta, reinforced by Scott Belsky. The bottleneck moves from production to curation — and curation is the irreducible human job.Layered Continuous MonitoringNo single data source is sufficient for population health inference. Combine financial claims, wearables, face scan biomarkers, labs, questionnaires, and partner integrations into a unified data lake — each layer compensates for gaps in the others.