A Summer of Vibes: How AI Automation Changed Our Team

How AI automation through Vibe Coding transformed my team

I love summer. I always have. It's when my birthday happens, it's when the sun shines in the northern hemisphere, it's when school's out, and for the Learning Experience team at BrainStorm, it's intern season.

As a Learning Experience Designer at BrainStorm, I’m responsible for creating scalable learning journeys for knowledge workers, specifically focused on technology adoption. As you can imagine, this landscape has changed dramatically as AI has come into its own.

Not only are we building courses about AI, we’re also using AI to manage our own processes. Designers, contractors, and interns work together to turn that vision into reality.

This summer, I expected our interns to keep the grunt work moving. Instead, they built things our company didn’t think were possible.

Here's What Actually Happened

There are a lot of manual, tedious, but necessary tasks when you manage a large portfolio of learning experiences like ours (we manage learning for 20+ apps and app suites). And historically, some of the best and brightest interns have been responsible for those processes.

This has included: video recording and transcription, digital asset management, content organization, image creation, event session creation, content tagging, auditing, and more.

This year, we asked our interns to not just manage those processes, but to transform them using vibe coding.

What is vibe coding?

A term coined by AI researcher Andrej Karpathy in 2025. It is the act of writing code by telling AI what you want in plain language. No actual coding required.

To be honest, I had my doubts about what they could do. The summer is short, and these interns ranged in experience with coding and AI.

I was happy to be proven wrong. Check it out:

What our interns vibe coded

Here’s what happened when we gave interns room to automate real work instead of simply maintaining it:

Our interns didn’t just augment our existing capacity. They automated new and existing workflows in ways that changed what the team could realistically accomplish.

Anson, the Monarch Intern

And, to close out with my favorite story, our most senior intern (we call him the monarch intern) came up with a plan to automate my second-to-least-favorite task: inputting surveys, assessments, and assets into the platform.

One of our developers said it was “impossible.”

Anson did it anyway.

Something that was manual, time-consuming, and accident-prone (copying from our design docs into the platform one question and response at a time), now happens with a few clicks. No human oversight required.

Why automation created more value than augmentation

These two terms are thrown around a lot in AI circles. The basic split is this:

Augmentation: Keeps the human actively collaborating

Automation: Only has a human intervene when needed. Think of a back-and-forth conversation compared to a set-it-and-forget-it task.

Anthropic recently published a study on how Claude is used, and a clear pattern emerged: users automating their work are getting the most value from AI. It made me reflect on my own role and my team. I’ve noticed this both in my individual role and how powerful automation is when applied at scale. Here's what it looks like in my role as a Learning Experience Designer.

Augmentation creates value

I’ve used AI in several time-consuming tasks to augment my work. It makes it faster, more accurate, and better overall when I’m involved at every step. But it keeps the same processes firmly in place.

Automation creates even more value

Limitations have always existed in Learning Experience (aka Instructional) Design. The software is old and clunky. The latest version of SCORM, the file type that most learning experiences are packaged in, for example, was created before some of our interns were born!

BrainStorm doesn’t rely on legacy tools or practices, and that’s why I decided to work here, but there were still things that I couldn’t do with my toolset and constraints.

That changed when I discovered vibe coding. Two things that weren't even possible for me before became reachable this summer:

1. New learning experiences

I designed experiences that couldn't be built in legacy tools. Realistic responsive interactions, branching feedback, AI tutoring by an AI tool, things that would have taken extensive development time in the past took me a few hours to take from start to finish.

Here's a module I created for a new vibe-coding course in our AI Productivity pack of learning experiences. Take a look if you’re interested!

2. Pie-in-the-sky apps

We collect a lot of data here at BrainStorm, with an over-90% survey completion rate for the AI packs, I could never keep up with all the data pouring in on my own. With vibe coding, I can write a script that gathers data from multiple sources weekly and surfaces it to me on a dashboard with easy-to-parse comparisons and flaggable takeaways. I now make changes based on data all the time, not just whenever I can dedicate half a week of work to it, and I cannot overstate how transformative that is for me.

What will your team vibe code?

The impossible becomes inevitable when non-coders get access to AI, the right tools, and encouragement to learn. At BrainStorm, we believe that shift—from 'Can we?' to 'How do we?'—is what separates organizations that use AI as another software tool from those fundamentally change their business.

The AI Productivity pack is built to enable this transformation. It combines adaptive training workflows, ready-to-launch campaigns, and actionable insights to help teams move from AI awareness to adoption and beyond to empowerment. Across the spectrum of hesitant to diving in too deep, the pack meets your people where they are and helps them build real skills in prompting, workflow integration, and identifying where AI adds the most value to their work.