Why Your Team is Probably Missing the AI Revolution (And NASA Can Explain Why)

Teams are at risk of missing the real AI revolution because most organizations are treating AI as an add-on to existing workflows rather than as a new “team member” that changes how cognition is shared. The core warning is that AI productivity gains at the individual level don’t automatically translate into team-level progress—especially when teams keep making decisions and managing context the same way they did before AI entered the room.

The NASA space shuttle story is used as a cautionary analogy: the ability to build the shuttle wasn’t preserved in any single person’s head. When the original teams disbanded and documentation scattered, the knowledge effectively disappeared—blueprints and specs weren’t enough. The crucial know-how lived in the collective connections among many people and countless small decisions. That framing sets up the central claim: AI is now capable of participating in that “between-heads” space, so teams must redesign their practices to distribute cognition across humans and AI.

According to the transcript, a divide is already emerging. Higher-performing product teams don’t just use AI to draft faster or generate ideas individually. They “distribute cognition” by building new team rituals and shared understanding around AI-generated work. That includes collective norms for prompts, explicit evaluation (eval) as a team rather than an individual afterthought, and workflow changes that reduce coordination overhead—tasks that used to require meetings can shift to AI-assisted coordination. These teams also rethink decision-making from the ground up, assuming AI is part of the team’s knowledge system rather than a separate tool.

By contrast, most teams are described as relying on subscriptions and chat-based generation—often taking AI output uncritically and even substituting AI chats for actual product requirements. The transcript emphasizes that the difference isn’t which model is used (ChatGPT, Grok, Gemini are mentioned as examples). The difference is cultural and procedural: how AI reinforces team norms, how shared context is handled, and whether the team treats AI output as something that must be integrated into product thinking.

A major practical requirement is managing shared context explicitly. Instead of treating documentation as a static “prompt bible,” teams should curate and feed key inputs to AI as part of the natural workflow—refined decisions, diverse inputs, and other context that the team deliberately maintains. Without that, AI can speed up individual output while weakening the team’s overall quality and alignment.

The transcript closes with a broader implication: AI is increasing optionality—sometimes by orders of magnitude—so teams should expect to iterate more and rethink processes accordingly. If organizations keep using AI to accelerate old patterns, they’ll underestimate AI’s potential. The challenge posed is direct: are teams using AI collectively as a form of shared intelligence, or merely using it to make individuals faster at the same old work? The answer determines whether AI becomes a true team advantage or just a faster way to produce misaligned results.