A research platform in progress, AI at a $2 billion bank, and academic research.
A research system for behavioral and social scientists, built by someone who has run the labs. The first module checks your IRB submission against your institution's ethics requirements before you send it in.
Next is a simulated pilot: AI agents take your study before real participants do, so you find the holes in the design before you spend money running it.
I lead AI strategy, adoption and efficiency at First United Bank & Trust, a bank with $2 billion in assets. I build with Power Automate, Power Apps, Excel Scripts and AI agents to take repetitive tasks off people's plates, cut what the bank spends on manual work, and free up leadership for the decisions that actually need a person.
My LSE dissertation asked how using AI at work affects employee well-being. Light and heavy users both reported lower subjective well-being than moderate users. Models have improved a lot since I submitted it, so the question is worth asking again.
At Harvard Business School I built an LLM chatbot and A/B-tested its response timing, grammar and tone. 77% of participants believed it was human.