The Offer Letter Brian Piotrowski

Modernizing TA Through Curiosity and Small Experiments

By · 4 min read

This whole thing started because I got curious, and remember talent acquisitions is a lot of pushing yourself. This took me a long time to figure out how I wanted to share this story.

Like I usually do, I gave myself a crash course into a new platform I learned about called vibe coding. I wanted to make something simple a tool that enriched publicly available data I was already collecting online. Similar to a program that aggregates data like how Indeed feeds it’s jobs, nothing fancy. Just something to see if I could do it without any programming experience. Well… I did take one Visual Basic class in college.

AI tools aren’t going to magically build a polished product for you. I never ended up creating a tool that successfully solved what I wanted.

But they will absolutely help you build more than you expect if you stay curious and keep experimenting.

This led me to a bigger question:

If I can (almost) build small pieces myself, how much of the recruiting tech stack already exists, and how much can be improved with the same mindset?

Turns out, a lot.

As I kept searching, I realized most TA teams already have the core pieces they need like an ATS. The problem we are trying to solve is more capability or connecting other tools.

Instead of buying another platform, sometimes it’s worth asking: Does something already exist that solves part of this? More often than not, the answer is yes. Open‑source tools, small products, and lightweight utilities can fill gaps without forcing you to overhaul anything.

Once you start exploring what’s out there, you stop thinking in terms of buying new tools and start thinking in terms of assembling what fits.

Tools like ChatGPT, Bolt.new, and VibeCoding make it easier to try things without depending on engineering or waiting for a vendor. You can literally describe what you want:

“When a candidate moves to interview stage, send a confirmation email and notify the hiring manager.”

And AI will help you generate the connector or workflow to try it.

This removes the barrier of needing a programmer to help you develop a solution and lets you design the workflow as the end user.

I already dislike large, bulky HCM products because they’re usually companies that acquired several smaller systems and bolted them together. What if you could have best‑in‑class solutions but build your tools to communicate internally? What if you could even build a solution internally?

Here are examples of things TA teams can build with curiosity, open tools, and small bits of AI‑generated code:

  • Summaries of interview recordings attached to candidate profiles.
  • Skills extracted automatically from resumes.
  • Alerts when a pipeline goes quiet.
  • Simple talent pools created from approved public data.
  • A tiny chatbot that helps candidates schedule interviews.
  • Offer reminders when something has been sitting too long.

If you want to start experimenting, you don’t need a major plan.

A simple approach:

  • Pick one small, annoying task.
  • Search GitHub for something similar.
  • Ask AI to help you build a connector.
  • Test it with dummy data.
  • Adjust based on what you learn.
  • Decide whether it’s worth integrating.

If you’re a user of an enterprise product, you can still explore open‑source and automated workflows even if you use Oracle, Workday, or SAP.

These systems all have APIs and ways to extend functionality. You can try new ideas, enrich data, or build dashboards then feed the results back into your main system safely.

Experimenting doesn’t mean ignoring responsibilities.

A few things to remember:

  • Be thoughtful about data privacy and consent.
  • Double‑check security expectations.
  • Look at how active an open‑source project is.
  • Make sure small tools can support your volume.
  • Know who will maintain what you build.
  • Always involve Legal and InfoSec before touching candidate data or automating anything that moves information.

The more I explored, most of this is possible if you have a willingness to explore what’s already out there because most of the tools, data, and building blocks already exist.