Three Patterns Reshaping How Organizations Adopt AI

Adopting AI as an individual is one thing. Adopting it as an organization is a completely different challenge. Companies like Block, Postman, and Anthropic are starting to figure it out, and three clear patterns are emerging from their announcements, interviews, and podcasts over the past few weeks.

1. Flatter Organizations

When centralized intelligence can distribute organizational knowledge across an entire company, you do not need as many layers of hierarchy to disseminate information. If your artifacts and institutional knowledge are centralized -- even in something as straightforward as Notion -- AI can process and distribute that information effectively. The middle management layer that existed primarily to pass information up and down the chain becomes less necessary.

2. Smaller Teams

AI is making individual contributors significantly more productive. Anthropic reports that what used to take 20 engineers now takes about five. But this creates a management problem: if 10 engineers are each producing what previously required two or three people, a single product manager cannot effectively guide or direct all of that output. The solution is breaking teams down further. Instead of one PM managing 10 engineers, you end up with one PM, two to four developers, and a designer. Everyone is more productive, quality stays high, and total output increases.

Before AI	After AI
1 PM per 10 engineers	1 PM per 2-4 engineers
Large teams, slower cycles	Small teams, fast iterations
Role-specific artifacts	Shared code access

3. Blurred Roles and Responsibilities

The traditional boundaries between engineers, designers, and product managers are dissolving. It no longer works to say engineers write code, designers make mockups in Figma, and PMs write documents in Notion or Linear.

Now everyone can write code. Everyone can open pull requests. Designers and PMs will open different kinds of PRs than engineers -- the engineers handle the complex architectural work -- but if a button needs to change, the designer can just change the button. No more cycling through artifact creation, approval workflows, and handoffs.

PRDs are dying. Instead of writing long documents describing exactly what a feature should do, product managers can open Claude Code or V0 and build a prototype. Code is clay now. Use it to mold your vision into something people can touch, interact with, and discuss. It creates a much higher level of collaboration and more productive conversations about what you are actually building.

Key Takeaways

• Flatter organizations reduce hierarchy because AI can distribute institutional knowledge directly
• Smaller teams are necessary because AI-amplified productivity overwhelms traditional management ratios
• Role boundaries between engineering, design, and product are dissolving as everyone gains the ability to write and ship code
• PRDs are being replaced by prototypes built with tools like Claude Code and V0
• These patterns are emerging simultaneously at companies like Block, Postman, and Anthropic

Resources

• Anthropic -- AI safety company behind Claude
• Block -- Financial technology company (formerly Square)
• Postman -- API development platform
• V0 by Vercel -- AI-powered UI generation tool