the ai sneak attack we’ve seen coming for decades

There’s a running Monty Python bit here that starts with this taunting

Today I’m building an artificial intelligence agent to solve a business challenge at work. That’s something I couldn’t have predicted a week ago.

Last month I was a Content Designer. This week, I’m thinking about titles like Prompt Engineer and Agentic Developer.

TBH, I’m not actually building the solution, I’m practicing building an AI assistant so I can understand how it works better. Monday I will bring the idea to my colleagues to see if it’s a viable approach to experiment with. It’s crazy how fast AI is already going, but it wasn’t unpredictable.

ai is changing every technology career – my accidental journey from content designer to conversation designer to agentic developer?

My journey toward AI started in 2016. I was at an HR tech startup, I launched an Intercom widget in our alpha product release to be able to get user feedback from our first group of users.

By 2018, I was designing conversational interfaces for Penny, Credit Karma’s chatbot serving millions of users a day. I was developing and shipping financial coaching conversations at scale and ingesting tens of thousands of inputs per day using a manual application of natural language processing developed by our engineering team.

A year later, I was working with data scientists to develop machine learning to run propensity models to serve up personalized content on Credit Karma’s front door at scale.

Today, I used the world’s most powerful Large Language Model to create a product launch strategy for a coffee startup, then moved on to developing an AI agent for a commercial banking product.

AI’s slow advance and sudden attack feels like the scene from Monty Python Quest for the Holy Grail wherein John Cleese’s Sir Lancelot runs over the horizon for ages, and then is suddenly upon us.

Tap image for the video metaphor of AI in 2024 that ties in all these weird images

the speed of solutions to real business problems is outpacing business processes

The focus of my current project is on improving support experiences while reducing support costs for a banking product. Our goal this quarter was to reduce the costs of filing user disputes, a highly regulated experience that often requires human involvement and is associated with fraud risk. When we set our OKRs for this quarter, we were excited to be using an intelligent voice assistant (IVA) product that helps users on hold for support self-solve their problem. Two months later, after shipping incremental UI improvements and the new IVA, we learned these approaches weren’t helping much.

Now, quite suddenly, we’ve found we have the AI tools to build an AI agent that can collect the initial information required for the user’s dispute claim and create the claim ticket through our internal systems. If human interaction is still required for risk or legal purposes, that smaller group of users can be connected to a live agent via telephone, SMS or chat.

If successful, after tuning and refining, we can safely expect this agent interaction to cut call volume at least in half and improve experiences by eliminating the wait time for expert help as needed. This alone more than pays for the technology.

eagerly imagining the future faster than ever before in human history

The future of work is so exciting right now. I can’t remember the last time I spent so much energy learning about a new technology, especially one I thought I was ready for. Although many technologists are preparing for an employment doomsday event, I find myself looking forward to the day that AI can do the parts of my job that don’t require imagination and emotional understanding, so I can be fully involved in the satisfying work of imagining a better future for people.