This post originally appeared on LinkedIn.

The latest Andreessen Horowitz-Mercury AI spending report revealed something fascinating about how enterprises are actually using AI. The data, on how Mercury’s 200,000 customers spent on AI tools, shows that creative tools and coding platforms have broken free from their departmental silos. When Replit (ranked #3) and Midjourney (ranked #28) spread across entire organizations - not just across engineering and design teams - we are witnessing specialized capabilities dissolving into universal skills.

The Vibe Coding Revolution Is Real

SaaStr founder Jason Lemkin recently highlighted the explosive growth we're seeing:

"Many talk about SaaS being dead. I think they miss that @Replit + other vibe coding leaders already have created 100,000+ new web SaaS apps. Already. In less than a year. Everyone is just getting started here. I've already built 8 web apps myself in 100 days. Up from 0 in the past 10 years."

This isn't just about professional developers anymore. As Lemkin notes, vibe coding platforms are enabling a "renaissance of 1 million new SaaS apps" by making it possible for anyone to build. The democratization is real, and it's happening faster than most realize.

The Hidden Challenge of AI Democratization

This democratization sounds like pure upside: finance teams are building applications, sales is automating workflows, and marketing is shipping code. Everyone can build now. But with great tools comes... great chaos.

McKinsey's recent report with Reid Hoffman on "Superagency" reveals another dimension: almost 90% of leaders anticipate AI will drive revenue growth in the next three years, but nearly 70% of transformations fail. The gap? Organizations are investing in AI tools but haven't built the infrastructure to capture and scale the innovations happening at the individual level.

Here's what's actually happening inside organizations today:

• That killer automation your sales rep built in Cursor (#6)? Only they know about it.

• The prompt your analyst uses to 10x their output? Buried in their ChatGPT (#1) history.

• That Replit workflow saving 5 hours per week? One person's secret weapon.

We are creating workflow innovations faster than we can capture them. Every day, employees across your organization are discovering breakthrough ways to use AI, and every day, those discoveries remain trapped with individuals.

The Real Competitive Advantage

The winners in this new era won't be companies with the most AI tools. According to the a16z data, 70% of the top AI applications allow individual adoption before team deployment. This bottom-up adoption pattern means innovations are happening everywhere, but without infrastructure to capture and scale these discoveries, organizations only realize a fraction of the potential value.

As Lemkin points out, monetizing these new capabilities "will be...different." The same applies within enterprises: capturing value from distributed innovation requires new approaches.

The companies that will dominate their industries are those that turn every employee's AI discovery into everyone's advantage.

Building Your AI Discovery Infrastructure

What we need is a "GitHub for AI workflows"; except this time, everyone from sales to finance can actually use it. This isn't about complex version control or code repositories. It's about making sharing as easy as discovering.

Here are five steps organizations can take to start building this infrastructure today:

1. Know What’s Actually Happening

Before you can capture innovations, you need to know where they are occurring. Tools that track AI application usage at the page level can reveal which teams are pioneering new workflows and where the real productivity gains are happening. Quick pulse surveys ("How did AI help you today?") can surface breakthrough use cases you'd never discover otherwise.

2. Create a Central Repository

Start simple with a shared space where teams can drop their AI wins - the prompts, the workflows, the automations. This could be as basic as a dedicated Slack channel or as sophisticated as a purpose-built platform.

3. Make It Accessible

Unlike traditional developer tools, your AI discovery system needs to work for everyone. Use plain language, visual workflows, and clear categorization that makes sense to non-technical users.

4. Incentivize Sharing

Recognition matters. Celebrate the sales rep who shares their Cursor automation. Highlight the analyst whose prompt technique gets adopted company-wide. Make sharing discoveries as rewarding as making them.

5. Enable Discovery at Scale

Tag and categorize contributions so teams can quickly find relevant workflows. A marketing manager should easily discover what the sales team learned about using AI for customer communications.

The Path Forward

AI tools have democratized skills - anyone can now build, create, and automate. But without infrastructure to democratize the resulting discoveries, we're leaving enormous value on the table.

Every company needs to evolve from asking "What AI tools should we adopt?" to "How do we capture and scale what our people are already discovering?"

The organizations that build this infrastructure won't just be more efficient - they will compound their AI advantages daily as every employee's breakthrough becomes organizational capability.

At Larridin, we are building the infrastructure that helps enterprises capture, scale, and govern these distributed AI innovations. Our Scout product provides visibility into AI adoption through a managed Chrome extension that tracks application usage at the page level, while enabling admins to gauge productivity impact with quick pulse surveys. Combined with our Nexus platform's compliance-ready prompt libraries and authorized workflow sharing capabilities, organizations can both discover where AI innovation is happening and scale those discoveries into organizational advantages.