How I’m Using My HR Experience To Build Better AI
Discussing my journey from leaving a career in Human Resources to learning how to leverage all that experience to utilize AI in a whole new way.
Until very recently, I was a successful HR Leader at several Fortune 500 companies. Then I chose to walk away. It is very strange to step away from something you are good at and know well. And it’s honestly intimidating because I don’t have any experience applying these specific skills outside of Corporate America. But I know I can. So I started thinking about what I did know before I tried to figure out what else I could do with it. And as I reflected over my 20-ish years in the industry, I gained a lot of experience that I’m genuinely proud of.
I have my Masters Degree in HR from the University of Illinois’ School of Labor and Employment Relations.
I started my career at Shell in various capacities as a part of their leadership development program, including sitting in on union negotiations, outsourcing the whole FMLA process for the entire US, and spending a couple months in Singapore as the HR Manager trying desperately not to (metaphorically) drown.
I moved from Houston to Chicago to work for Abbvie, a biopharmaceutical company. I supported their global marketing teams that were working on selling a wide array of drugs across the world. Treatments for RSV and HIV. Curing Hepatitis C. Developing and unlocking cancer fighting drugs. It was the first time I really learned about marketing and global economies.
I then worked at JLL, a massive commercial real estate firm. There I supported Life Sciences accounts in the US, then various organizations (Sales, Procurement, Product) across the world. Again, it was a great opportunity to learn new skills, as we had to build organizations from scratch (Product), globalize functions (Sales, Procurement), and create teams, systems, and training to maximize the talent, for our people and clients. It was also the first time I had a proper direct report, so I had to learn to be a “boss” in real time.
Finally I moved over to Gallagher, a large global insurance broker. I worked in their Benefits, Wholesale, and Specialty divisions. I had a large team across multiple countries. I supported leaders that were growing revenues, expanding organizations to unlock greater value, and really understood how decisions were made and where my skills best supported the business.
I share all that not as an extended resume, but because it’s important to remember and articulate what I was a part of creating, building, operationalizing, and executing throughout my career, from both an HR and business standpoint.
When I walked away from Corporate to try and build something of my own (one by myself and the other with a friend), I said explicitly that I wasn’t going to do anything close to HR. I wanted to try something completely different. I wanted to challenge myself, and honestly give myself a break. It’s a draining career working with and helping people constantly. Rewarding, but it can be a lot. So I said I’d pivot and go in a new direction.
I wasn’t sure what it would be (still unsure, but hopeful), but I knew that I needed to get up to speed on AI to boost my odds and my abilities. So I started to experiment with it. Put in some queries. Learned what a “prompt” was. Listened to podcasts I’d never heard of, read articles on topics I’d never considered..
As I played with the technology, I’d stumble upon more and more. Vibe-coding in Replit to build a basic app prototype instead of what I’d normally do to explain a complex concept…PowerPoint slides! I’d see that this “agent” word kept popping up. Mess around with image generation and song writing. All to just see what the capabilities were with the AI, and with my ability to leverage it.
And the AI kept improving, at rapid speeds. Google stepped up to the plate and started to release some really significant things. Not just Gemini, but Nano Banana and Flow and Antigravity. And as a result, you saw Claude and ChatGPT and others step up too. Concepts and industries were flourishing. But it was a lot. I didn’t know how to leverage it. I didn’t know how to organize it. That’s when I stumbled upon something. I would turn back to what I knew. I’d go back to my HR roots.
In my very simple terms, AI tools, chatbots, agents, etc. are all just responsible for specific tasks, in specific ways. Chatbots and LLMs generate answers based on the probability of the next word in the string (based on a billion lines of code and logic). Agents were just focused, diligent, narrow versions of general AI chatbots, trained on very specific tasks or topics. I viewed it like an orchestra. Each tool, each individual chat, was an instrument. Let them all play whatever they wanted at once, or without clear direction, it would be a noisy, cacophonous mess. But give them a purpose. A song to play. With a conductor to help them know when, how fast, how loud…then you get a symphony.
It made me think of AI, and that made me think of HR. No, I’m not comparing an HR professional to Beethoven or Bach. But I am saying HR leaders (and business leaders in general) are constantly looking at the collection of people, of talent, that they have at their disposal and it’s their job to maximize their abilities together. So I have experience in skill in that area. And it’s what effective AI needs. To have someone helping all those tools work together towards a common goal or output. You need a conductor. You need a leader.
So I started to learn and apply things in familiar HR or Business terms. For instance:
How could I get my AI Chatbot (Claude) up to speed on everything I’d been working on for months?
I had to think of Claude like a person. A really skilled, really capable person that had no detail or context in the specific area of interest that I had. If I had that person in front of me, I wouldn’t just dump dozens of documents on their lap and then assume they’d understand it all and connect all the dots automatically. I’d expect them to read it. To understand it. To bring in their personal knowledge. But it would be a really ineffective and unrealistic approach if I didn’t also make the time to explain. To add context and nuance and texture to the words on the page. I’d have to invest time. I’d have to ask questions, and answer ones back. That’s how you do it right. So that’s what I did with Claude. It took a lot of time and energy, and it “surprised” Claude. But when I started to approach it as a new hire to onboard to the best of my ability, the responses became much, much better, and more aligned with my thinking.
How do I use AI without having to prompt and explain the same things over and over again?
Well, you need to do three things:
Define the exact inputs you have and outputs you want (templates and workflows)
Articulate why you need those outputs and what you’ll do with them (context matters)
Imagine you have an unlimited number of Harvard MBA hires that will do whatever you teach them to do on a computer (and build a smart, specific plan to do that teaching)
To me, that was the basics of building automated AI workflows. You define the trigger event. You walk through the specific steps and actions to arrive at an output. You make sure the inputs, processing, and outputs are the format you want. And then you get that output, review it, and teach the AI what it got right, wrong, and halfway.
Once you’ve built it, you make sure you test it. You check it. And you make sure you have a “canonical” source document. I call this the Source of Truth document. Process documentation becomes absolutely essential. Because if it’s written or recorded, it can be referenced and learned. If it’s sitting around in your brain, no one and no thing can access it.
This is fundamental learning and development work. This is textbook org design. This is SOPs and process mapping. This is HR.
The employee base are AI agents instead of people. But the fundamental principles of onboarding, org design, L&D all apply.
What’s an example of an automated workflow?
Well, I use my travel agency (Rewire Travel) as a sandbox for things. So I had someone reach out and say “I want to take my wife to Paris” for the weekend, can you help? I need A, B, and C.
What I’d typically do is ask a lot more questions and start to research. Read blogs and TripAdvisor reviews. Ask ChatGPT for ideas based on the simple prompt. But that takes a lot of time and is still limited. So I built a workflow based on the steps. What steps do I really take to help someone plan a trip?
Let’s start with the hotel. First, I ask the client a bunch of questions (budget, amenities, dates, etc.). Then, instead of doing all that searching myself, I now ask Claude to use the prompt I wrote to do the first round of research. That prompt has both the client’s responses and what I as the expert know is critical to make the best recommendations. The prompt is fed into Claude to do deep research. The resulting outputs are formatted in they way I like, because I wrote the template that Claude has to use. Then I personally review them. Once my review is complete, I make my final recommendations and send them to the client (via the template I built that Claude populates). The client has a clean, curated proposal. It has my stamp of approval and my decisions. My method. My tools. But I delegated the first round of research to my employee the way I taught them to do. That employee just happens to be Claude.
Each of those steps, besides the human inputs of hotel needs and review, are AI agents. If I ran a travel agency with staff, I’d still sit with each of them, outline what they need to do, how they need to do it, what I need, what the client needs, and more. It’s the same effort. Same explanations. Same thinking. But it’s much faster, much easier, and much more adjustable with AI agents. They aren’t just search engines. They aren’t dumb, unremembering entities. It is incumbent upon you to teach them and build a framework. If you have that, they can do damn near anything.
So that’s what I’m starting to explore. That’s what I’m leveraging as a former HR professional. I’m stepping back and thinking about the organization and the talent I need to get the outputs and results I want. You have to define and scope the role. You have to select for certain skills or experience. You need to onboard, and develop, and coordinate. There are so many aspects of HR that are applicable to being really effective and successful with AI, particularly agentic AI. So I’m going to keep experimenting and learning. But now, I’m not pushing my HR skills away. I’m bringing them to the forefront.
