Elevating your AI game (with a day job)

I was thinking recently about how to best integrate new AI tools into our lives as cloud security professionals.

And I concluded: all hours of the day aren’t created equal for this. Let me explain.

As working professionals, many of us operate within the structure and scale of large enterprise environments.

These organizations are complex, often carrying the weight of legacy systems, strict regulatory requirements, and deeply embedded processes.

While these constraints are necessary for stability and compliance, they also mean that the adoption of new AI technologies tends to lag behind what is available in the broader tech ecosystem.

This isn't a critique of enterprises. It’s just the nature of the beast.

But there's a growing disconnect between the tools available to us in our professional roles and those we can access outside of work.

When we step outside the enterprise environment, we enter a different reality.

A reality where you can vibe code a CRM solution in 15 minutes with bolt.new, Replit, or Lovable.

Given this contrast, one useful mental model is to categorize our career activities into two buckets:

1. "Day Job" Career Activities: tasks performed within the enterprise environment, on corporate-managed devices. Innovation here is incremental, and tool adoption is slower.

2. "Outside of Day Job" Career Activities: self-directed initiatives we pursue in our own time and space.

It's the second category that presents the richest soil for experimentation and rapid skill development.

The lack of constraints allows us to test new tools, explore novel workflows, and stay ahead of the curve—sometimes by a year or more.

So embrace the dual reality. Use your evenings, weekends, and passion projects as a sandbox to play with the future—because by the time it reaches your day job, you’ll already know how to make it work.

Things I've Been Enjoying

GPT4o's new image generation mode

This has been SO much fun. Revolutionary stuff. Like the rest of the internet, I've become addicted to Studio Ghibli-fying images:

This one deserves it's own post - added to my backlog.

Gemini 2.5 Pro Experimental

New state-of-the-art model across most benchmarks (topping LMArena leaderboard). This model really shines with long context uses cases (can handle 1M+ tokens!!).

Another one that deserves it's own post - I'd love to once I've played around with it more.

A16z's Top 100 GenAI Consumer Apps (4th Edition)

This is a list of highest traffic / most used AI consumer apps by venture capital firm Andreessen Horowitz.

Fascinating to see what makes the list and how it has changed since previous iterations. Few funny ones in there, including the "LooksMax AI" app which has been downloaded over 5 million times:

Let me know if you want me to try this one for the good of science.

Inside Bolt: From near-death to ~$40m ARR in 5 months (Lenny’s Podcast)

Very enjoyable episode covering the founding story behind Bolt.new.

The main takeaway for me was how critical their "WebContainer" technology was to the success of their platform. So really, it was an "overnight success 7 years in the making".

Quote of the day

 “The future is already here – it's just not evenly distributed.”
— William Gibson

Hope you got something from this braindump.

Till next time,

Nelson