Most AI pilots fail due to poor change management and a lack of business context. A successful model involves embedding vendor engineers within the client's team to handle agent onboarding, systems integration, and process customization, ensuring the AI works within the company's unique environment.

The traditional, linear handoff from product (PRDs) to design to dev is too slow for AI's rapid iteration cycles. Leading companies merge these roles into smaller, senior teams where design and product deliver functional prototypes directly to engineering, collapsing the feedback loop and accelerating development.

Hanover Park’s organizational design has no product managers or designers. Instead, it embeds engineers directly with fund accountants—the domain experts. This creates a tight feedback loop that allows them to build a more informed and practical product much faster by aligning development directly with user needs.

The process of building a custom AI agent forced Newell's teams to collaborate more closely than in traditional software rollouts. It sparked critical conversations about existing versus ideal workflows, bringing people together to solve problems and improving organizational connectivity as a positive side effect.

Instead of randomly applying AI, a better approach is to journey map the internal process of how product, design, and development teams collaborate. This analysis reveals the biggest bottlenecks and points of friction, which then become the most valuable and targeted places to apply AI for genuine process improvement.

Brex structures its AI teams into small pods, combining young, AI-native talent who think differently with experienced staff engineers who understand the existing codebase, product, and customer needs. This blends novel approaches with practical execution.

A new organizational model is emerging where companies create small, agile teams comprising a senior expert, an engineer, and a marketer. Empowered by AI tools, these pods can develop and launch new products in a week, a task that once required large teams and over six months.

The traditional product workflow—writing PRDs, waiting for mocks, then building a prototype—is being collapsed by agentic tools. A single "Builder PM" can now perform user research, generate PRDs, create functional mocks, and build a working prototype, drastically shortening the feedback loop.

Commure adapts Palantir's model, embedding young engineers directly within hospitals. These engineers work alongside physicians to co-develop and iterate on AI models in real-world settings. This on-the-ground presence accelerates adoption, builds trust, and ensures the tools solve real clinical problems.