How I Made AI My Unfair Career Advantage
How to Spot Opportunities Early and Turn Them Into Your Leverage
There’s a question my boss started asking me more and more often over the past year. At first, it happened every few weeks. Then weekly. Now it’s almost every conversation.
Jenny, how did you even do that?
Each time I’d show him something new. An autonomous testing workflow, instant SQL queries from a database with hundreds of tables, a research system that synthesized reports in minutes instead of days… Every time he’d pause, shake his head slightly, and ask that same question.
At first, I thought it was just curiosity. Maybe even a compliment.
But looking back, I realize what he was really asking was: “What changed? How did you go from predictable engineer to someone who keeps solving impossible problems?”
That gap between his confusion and my reality? That’s the story of how AI completely transformed my career in a traditional biotech company. And it represents the biggest opportunity hiding in plain sight for anyone working in any industry right now.
Part 1: The Traditional Path That Almost Made Me Invisible
The PhD Who Disappeared Into Competence
Let me set the scene for where I was before all this started.
I had what most people would call a good career: a PhD in hand, a software engineering role at a biotech company, comfortable salary. The kind of job that checks all the boxes on paper.
But nobody tells you about following the “smart path”: it can be mind-numbingly predictable.
My work was solid. My code was okay. My projects shipped on time. And I was completely invisible in terms of innovation potential.
Weekly one-on-ones with my boss followed the same script: project status, technical details, maybe a quick discussion about team dynamics. I’d take notes, nod along, and leave knowing next week would look exactly the same.
I wasn’t unhappy. I was just… there. Reliable but unremarkable.
And honestly? That felt safe enough.
When Curiosity Started Changing Things
The shift didn’t happen overnight. It started with small curiosities.
I’d been using ChatGPT for basic tasks — drafting emails, summarizing documents, the usual productivity hacks everyone was trying. Nothing groundbreaking.
But watching AI in action, I noticed it could provide guidance, follow instructions, and make decisions based on context. If you gave it the right tools, it could collect information autonomously and participate in decision-making loops.
That changed how I looked at problems at work.
Take our business development team. They were drowning in research requests. They needed information on dozens of preclinical drugs: clinical trials, market projections, company pipelines, patient enrollment numbers. The kind of work that meant hours of manual Googling, opening dozens of tabs, copy-pasting into spreadsheets.
I watched them struggle with this and thought: Could AI handle this?
Not just advise. Actually handle it.
Part 2: The Transformation Through Small Experiments
Project 1: The Drug Research Breakthrough
I decided to test something with Cursor. I took one drug from their research list and threw it into Cursor with five specific questions:
What are the available technical details?
Which companies are developing this drug?
What clinical trials have been conducted?
How many patients have been enrolled?
What’s the market value and projection?
What happened next genuinely surprised me.
Cursor didn’t just search. It orchestrated. It pulled information from clinical trial databases, SEC filings, research papers. It synthesized everything into organized output. What normally took our team hours per drug happened in minutes.
My coworkers were impressed with that first version. But I kept pushing.
I built it into a complete system with two AI agents: a Planner that designed research strategies and detected gaps, and an Executor that fetched data and kept everything organized. The research that used to take days now happens autonomously.
That was the moment I realized AI had fundamentally shifted the playground. On the surface, it looked like a speed boost. But dig deeper, and things that were previously impossible suddenly become achievable. That magic clicked for me.
And more importantly, nobody else at the company was building things like this yet.
Project 2: The Autonomous Testing Demo That Shocked Everyone
Around this time, I’d been exploring how far I could push Cursor’s autonomous capabilities. I was working on UI performance testing, that is something our team had tried before but never got working consistently.
With Cursor, I managed to get autonomous testing actually running. Real tests, generating themselves, executing, catching issues.
When I showed my boss the chat history of how Cursor was autonomously testing our UI, he was blown away. He immediately asked me to demo it in our group meeting.
I agreed, not quite realizing what that demo would trigger.
When I walked through how the autonomous testing worked in front of the entire team, the room went quiet. Everyone was giving me these shocked looks. Because truly, a few months ago, this level of automation was impossible. We had tried and failed.
And now it just… worked.
That meeting changed how people saw me. I wasn’t just the boring engineer anymore. I was the person building things that looked like magic.
Project 3: The Database MCP That Changed Everything
Then came the project that really accelerated everything.
Our company database is massive. Over 400 tables. Some tables have billions of records. Triggers, sequences, indexes everywhere. It’s unwieldy and honestly a nightmare to navigate.
Before AI, every time someone asked me for data, I’d have to stop everything. “Hey, can you pull this report?” or “Can you fix this record in the database?” meant digging through the schema, reconstructing relationships, and hand-writing SQL queries.
Each request took 30 minutes minimum. And they happened constantly.
So I created personal tools for Cursor that could access our database schema and generate SQL queries on the fly. It resolved my pain points immediately.
I took a screen recording showing how Cursor could now generate SQL queries and autonomously plot the output. When I showed my boss, he was genuinely shocked.
Then, MCP became more widely known. My boss quickly thought of me and tentatively asked, “Could we build our own version for the company?”
I could tell he had minimal hope. This sounded ambitious.
I spent an hour one afternoon exploring it. Within that hour, I proved it was possible and had a basic MCP connecting to Claude's desktop.
While I can’t show the company version of that MCP, I’ve built a Substack Newsletter MCP that follows the same principle.
That proof-of-concept was enough to allow me to build it out properly. Finally, we had an MCP that connected our database directly to Claude through natural language.
Now, when someone asks me for data, I’d just paste their request into Claude. And it would return the complex SQL immediately.
The first time I used it for a colleague’s request, they replied: “It works like a charm.”
I responded: “Thanks to Claude.”
They had no idea what that meant.
But I did. And that knowledge gap was becoming my advantage.
Part 3: The Ripple Effects Nobody Expected
The Management Meeting Where Everything Shifted
In the past, top-level management meetings were pretty standard. Boring, even. Progress reports, status updates, predictable plans.
This time, my boss asked me to take part in presenting the database MCP we’d built.
I knew what he was thinking. He wanted excitement, challenge, something that could break the boring rhythm and potentially have a huge impact down the line.
I walked in expecting questions, maybe some skepticism.
What I got instead was silence.
The entire room, with VPs, directors, and senior leadership, all sat there quietly after the demo. They were processing. Because they’d never experienced this level of autonomous working before. It was completely outside their reference frame.
After that meeting, things got interesting fast.
When People Started Skipping Meetings to Install AI Tools
Within days, people started reaching out to me. Not just casual questions — urgent requests.
“Can you help me install this thing? I have time right now.”
I’d think: Don’t you have a meeting to go to?
Their response: “That’s not important. I want to try this MCP.”
That’s when I realized the change was happening. AI had moved from “interesting concept” to “I need this working on my machine immediately.”
Even VPs started using the MCP and sharing their vision of the future in our company-wide Slack channels: sipping coffee, reading newsletters, and making high-level decisions while AI handled the complex data work.
People don’t naturally welcome AI. Before my presentation, many of them were skeptical. They thought we were just talking about AI, not truly making it work.
But once they experienced it, once they saw it working, expectations changed overnight.
The Company-Wide Priority Shift
The most significant change came from my boss.
He came back to me wanting to double down. Not just on the database MCP, but on building an API MCP, and pushing for company-wide adoption of Claude instead of just Microsoft Copilot for enterprise.
This wasn’t a small shift. This was AI becoming the top priority for our entire technical strategy.
And I knew I’d played a real role in making that happen.
The ripple effects kept spreading. Leaders who’d been skeptical weeks earlier started DMing me, asking if I could educate their teams on Claude usage for specific projects. HR reached out wanting me to organize company-wide AI education sessions. More individuals kept reaching out, hoping for a quick chat to understand my vision and see how AI could help with their own workflows.
The Meta Moment: MCP Building Itself
Here’s the moment that really crystallized everything for me.
I was debugging the database MCP, trying to fix some edge cases and improve performance. And I realized: I was using the MCP itself to debug the MCP.
It reminded me of something I’d read: 90% of Claude Code is built by Claude Code itself.
I was living that same reality. About 80% of my MCP was built by the MCP itself.
That’s when the full picture clicked. This wasn’t just about productivity or automation. This was about fundamentally changing how we build, debug, and evolve our tools.
We were entering a new era where the tools help create themselves.
Part 4: The Unfair Advantage (And How You Can Build Your Own)
The Gap Between Corporate and Creator Worlds
Looking back, I wouldn’t have done any of this if I weren’t building in public. If I weren’t sharing my experiments and launches through my newsletter. If I weren’t connecting with people in the creator world.
That’s when it became crystal clear to me: there are two completely different worlds operating at different speeds right now.
In the creator world, where I documented my building journey and followed inspiring, innovative builders, AI evolution happens at a breakneck pace. Every day, someone’s launching a new tool, sharing a breakthrough, discovering something that didn’t exist last week.
That constant stream of inspiration pushed me to explore more and experiment harder.
But in corporate environments? It’s much, much slower.
Most of my colleagues weren’t actively using AI beyond occasional ChatGPT queries. When AI came up in meetings, it was usually abstract, future-tense: “Maybe someday we’ll integrate this.”
That gap was massive. While they were debating, I was building autonomous systems, creating custom MCPs, shipping solutions that slowly made the impossible possible.
That gap is the unfair advantage.
Why Being Barely AI-Competent Makes You Valuable
You don’t need to be a supercharged AI expert like the ones in the creator world to become invaluable in your workspace. That applies whether you’re in a traditional company or a tech-savvy one where most people stay comfortable.
You just need to be willing to experiment while everyone else waits for permission.
I wasn’t doing anything revolutionary. I was just:
Taking friction points seriously instead of accepting them
Building one small solutions, often after hours
Sharing what worked instead of keeping it private
Iterating when things didn’t work perfectly
Those big visible projects all started as tiny experiments. And there are dozens of smaller friction points I keep resolving alongside my work, tucked within my Cursor projects.
You can do the same.
Keep doing it for three months, and you’ll be surprised how much you can break through. When you keep providing small, tangible solutions, they start to stack up. At some point, the impact crosses a threshold that leadership can’t ignore.
Especially in environments where most people are still skeptical or waiting for official “AI adoption” plans, even moderate AI competence creates enormous leverage.
That’s how I went from a predictable engineer to the person leadership called for every strategic AI conversation. The compounding effect of small, consistent experiments is what makes you invaluable in the long run.
Your Move
This opportunity exists in your organization right now.
You’ve seen how it starts: one small friction point, one curious experiment, one thing that works better than before.
At first, people will just be curious. Then, they’ll start asking how you did it.
That’s the change between what they think is possible and what you’ve already made real.
While everyone else is waiting for “official strategy,” you’ll already be shaping it.
So ask yourself:
What friction point could you turn into your advantage?
What manual process could become autonomous?
What impossible thing could suddenly become doable?
The tools are here. The timing couldn’t be better.
The only question left is: Will you start experimenting now?
If this resonates, I’d love to hear what you’re building. What’s the first friction point you’re planning to tackle? Drop a comment or reach out. I read everything and genuinely enjoy connecting with people, bringing AI into every aspect of life.
Author Spotlight
We’re so grateful to Jenny Ouyang for allowing us to share her story here on Code Like A Girl. If you enjoyed this piece, we encourage you to visit her publication and subscribe to support her work!
Join Code Like a Girl on Substack
We publish 3 times a week, bringing you:
Technical deep-dives and tutorials from women and non-binary technologists
Personal stories of resilience, bias, breakthroughs, and growth in tech
Actionable insights on leadership, equity, and the future of work
Since 2016, Code Like a Girl has amplified over 1,100 writers and built a thriving global community of readers. What makes this space different is that you’re not just reading stories, you’re joining a community of women in tech who are navigating the same challenges, asking the same questions, and celebrating the same wins.
Subscribe for free to get our stories, or become a paid subscriber to directly support this work and help us continue amplifying the voices of women and non-binary folks in tech.
 | A guest post by
|
You captured the turning point perfectly. In my experience, all it takes is one working demo. The minute people see a real output that is actually useful they didn’t expect, scepticism is gone and momentum takes over. Nobody wants to be left behind once they realise how accessible this actually is.🩷🦩
Really nice work, Jenny! Staying curious and experimenting - that’s what actually creates influence and impact.