What It Really Takes to Deploy AI Across an Organization

Everyone wants to be an AI-native company. Of course, that’s way easier said than done.

Within companies, there are individuals being tasked with figuring out how to make their companies use AI effectively and to ensure that their AI pilot programs lead to real business value. These people — their companies' AI Champions — have a critical mission. But they lack playbooks for how to make that actually happen.

That's why we partnered with three other leading companies in the AI space — Cursor, Vercel, and Anthropic — to host an exclusive series of live workshops, all about what is currently possible with AI tools, how to run AI pilots that don't fail, and real stories of how the best companies are successfully integrating AI into their workflows.

Major themes and trends across all sessions

Across all four sessions, several key themes emerged that defined the most successful AI adoption initiatives:

• The Death of Traditional Development Workflows: AI fundamentally changes how work gets done, and development takes place at a more abstract level. For example, rather than engineers spending their time being hands-on-keyboard writing code, they start with a thinking model to generate a plan, iterate on that plan, then pass the plan off to an agent.
  ‍
• AI-First Organizational Design: The most successful companies are redesigning their entire operations around AI capabilities, rather than just adding AI tools into their pre-AI workflows.
  ‍
• The Prototype Revolution: Multiple sessions emphasized the shift from document-heavy planning to prototype-first development.
  ‍
• Community-Driven Adoption: The most successful AI implementations are driven by internal communities and champions rather than top-down mandates. Companies that can create an internal sense of urgency by spreading "AI FOMO" see more viral adoption across their organizations.
  ‍
• Beyond Basic Prompting: All sessions emphasized moving past simple AI use cases like summarization to sophisticated workflows. The real value comes from integrating AI deeply into existing business processes and using it as a thought partner, rather than just an assistant.

Cursor: Making an Engineering Org AI-native

Top takeaways

• AI enables new organizational structures: Expect flatter organizational structures with smaller, more autonomous teams.
  ‍‍
• AI coding tools empower non-engineers: Now, anyone can contribute directly to codebases.

At Cursor, the world's leading AI coding platform, they’ve seen how individual developers can transform their workflows. But it goes beyond individual productivity: the most transformative aspect of AI comes from enabling entirely new organizational structures.

They’ve even seen this within Cursor, as they’ve adopted a more flat organizational structure. "We have very small teams, and teams as small as a single engineer will own a feature from idea to marketing," says Nick Miller, a software engineer at Cursor. Perhaps most significantly, Cursor enables non-engineers to contribute directly to codebases: "Technical support engineers... can just ask the codebase. The same is true with go-to-market, the same is true with user ops. We all have access to a monorepo, and we're able to get information and build."

Vercel: Encouraging Company-Wide Prototyping with AI

Top takeaways

• Design systems can now be applied at massive scale: Tools like v0 ensure consistent branding.‍
  ‍
• Prototypes over PRDs: Instead of going through multiple rounds of stakeholder alignment and PRD development, anyone can now go straight to prototyping.

AI-powered prototyping tools are fundamentally changing the product development lifecycle, enabling faster iteration and broader participation in the building process.

Traditional product development starts with lengthy PRDs and multiple rounds of stakeholder alignment. But AI enables a completely, prototype-first philosophy: "Building our prototype with v0 is probably faster than how long it would take you to write a full PRD with every user flow and every feature spec that you actually want to try to ship," says Ary Khandelwal, Product Manager at v0.

That means that what used to take weeks can now happen in minutes. In the session, the Vercel team was able to demonstrate building a complete signup page in under 45 seconds. AI also ensures that design systems can scale, enforcing consistent branding across all prototypes through design systems that can be applied instantly.

Gumloop: Building Enterprise Workflows and Spreading AI FOMO

Top takeaways

• Start with workflows instead of agents: Your agents are only as good as the workflows they enable.
  ‍
• Create viral internal momentum by spreading AI FOMO: Successful AI adoption requires internally motivated employees, and you can motivate them with FOMO.‍
  ‍
• Drive adoption with well-designed hackathons: The best hackathons require extensive preparation before and promotion after.

Max Brodeur-Urbas, CEO of Gumloop, shared our playbook for viral AI adoption within organizations, including detailed workflows they use to run their own company and proven strategies for successful enterprise AI implementations. One key insight? Reliable workflows must come before autonomous agents.

"An agent is only as good as the tools it can use or the workflows that it can actually trigger,” says Brodeur-Urbas. “You can't just go straight to agents. It's like trying to be a quarterback without ever studying the playbook."

Successfully identifying and implementing reliable workflows is only the first step. Teams also need to make their AI successes highly visible. This visibility creates viral internal momentum through "AI FOMO." But you can’t create AI FOMO with top-down mandates.

What does work as a reliable method for driving company-wide adoption is well-designed hackathons. The best hackathons require extensive pre-education, the hackathon event itself, and aggressive post-hackathon amplification of successes.

Anthropic: The Playbook for AI Pilots that Stick

Top takeaways

• Go beyond AI 101: Don’t teach generic AI how-tos; explain how AI training can help your teams to their jobs specifically.‍
  ‍
• Select the correct success metrics: Identifying specific, measurable successes sets the stage for more future successes.

Anthropic has seen hundreds of enterprise AI pilots — which is why they’ve been able to identify both common failure modes and the specific strategies that lead to successful, lasting AI implementations.

The majority of AI pilots fail due to predictable issues: tools that aren't fit for purpose, poor pilot design and execution, or inadequate measurement strategies. Successful pilots must go beyond basic enablement and elementary AI training.

"Don't teach too much of the basics,” says Hannah Moran from the Applied AI team at Anthropic. “What you want to do in the pilot is help them understand how to actually do their job differently, and better." 

Once an AI pilot program has been implemented, how do you know if it succeeded? Anthropic’s team recommends focusing on specific metrics, not just generic metrics. One engineering team achieved "a 50% time reduction in ticket resolution" by having their AI agent handle initial ticket triage and information gathering rather than trying to solve bugs end-to-end. Identifying specific, measurable successes sets the stage for more future successes.

Implementing the AI Champion Playbook

Successfully becoming an AI-native company isn’t about what tools you adopt; it’s about fundamentally reimagining how work gets done. The most successful organizations are moving from document-driven development to prototype-first building, from hierarchical innovation to democratized creation, and from individual productivity gains to organizational transformation.

The most striking trend across all sessions was the speed of change. What took weeks now happens in minutes, what required specialized technical skills is now accessible to anyone, and what once needed large teams can now be accomplished by individuals equipped with the right AI tools.

For AI Champions within organizations, the playbook is clear: invest in proper enablement that goes beyond basics, create structured opportunities for exploration and learning, make successes visible to create viral adoption, and measure impact in ways that demonstrate real business value.

‍