How to build products on a moving frontier | Dan Shipper (Every)

Episode Overview

• This episode features Dan Shipper, Co-Founder and CEO of Every, sharing insights on how to build products on a moving frontier, particularly in the rapidly evolving landscape of AI.
• The talk addresses the inherent tension between executing a planned roadmap and exploring new technological possibilities, offering a practical organizational framework to balance both.
• Dan introduces the concept of a "Labs" team within a product organization to isolate exploration from execution and outlines how to manage this dynamic successfully.
• This content is highly relevant to product leaders, startup founders, and engineers struggling to integrate cutting-edge AI capabilities into their existing product lines.

Key Concepts

• The Exploration-Execution Tension: Product organizations face a fundamental conflict when technology moves quickly. Exploration requires divergent thinking, trial and error, and a high tolerance for throwing work away. Execution (scaling and maintaining) requires convergent thinking, focus, and a low tolerance for wasted effort. Trying to have the same people do both simultaneously leads to organizational friction and distraction.
• The Labs Team Solution: To resolve this tension, companies should separate concerns by creating a dedicated "Labs" team (which can be as small as one person). The Labs team's sole responsibility is to explore the frontier of new models and capabilities, with the explicit expectation that they will discard 90% of what they build. This protects the core product team, whose job is to scale the 10% of ideas that actually prove viable.
• The "Pirate and Architect" Archetype: A successful small Labs team benefits from pairing two distinct personalities: the "Pirate," a scrappy builder who write "vibe code" to quickly find value and proof of concept, and the "Architect," a systems-oriented engineer who takes messy, successful experiments and shapes them into clean, extensible, and scalable systems.
• The Research Pipeline: To move ideas from speculative Labs experiments to the core product, organizations need a structured pipeline: starting with lab-only experiments, progressing to internal use/dogfooding, moving to early customer testing, and finally handing off to the product team when the concept is ready to scale.

Quotes

• At 2:12 - "It feels like it's everyone's job to execute the roadmap at a high level and to stay at the frontier at the same time. And that's really, really hard because these are very opposed ways of working." - This explains the core organizational dilemma that Dan's framework aims to solve.
• At 5:49 - "On a Labs team, your job is to explore capabilities, run a ton of experiments in parallel... and expect to dispose of 90% of what you make. A product team is very different: your job is to improve and scale... and expect to adopt about 10% of what the Labs team makes." - This quote clearly defines the differing expectations and metrics of success for the two separate arms of the organization.
• At 8:39 - "The biggest thing that matters on a Labs team is making the feedback loop as tight as possible between making something and knowing if it's good... and the tightest feedback loop is making something for yourself." - This highlights the critical importance of dogfooding and rapid iteration in the early stages of frontier product development.

Takeaways

• Establish a dedicated "Labs" function—even if it is just a single engineer with protected time—to explore new AI models and APIs without disrupting your core product roadmap.
• Run multiple competing experiments in parallel to solve the same problem; because the technological frontier is highly uncertain, having different team members approach a problem from different angles is the fastest way to map out what actually works.
• Implement a structured "Research Pipeline" with clear stage gates (Lab-only -> Internal dogfooding -> Early adopter testing -> Product handoff) and use retention/usage metrics rather than initial novelty to decide which experiments graduate to the core product.