Hackathon Winners Part 5: AI-assisted Knowledge Chatbot Helps Find Content

Meet the team who landed two awards.

Hackathon Winners Part 5: AI-assisted Knowledge Chatbot Helps Find Content

Our AI theme continues with the penultimate feature in our six-part Hackathon winner series, bringing our total to four teams who have leveraged the power of this ever-evolving tech.

Related Articles: Meet Our Hackathon Winners

Meet Bertie Bott’s Bit Bandits, one of the three teams who secured the coveted Quaffle Award — better known as the most impactful end-user experience award for BigCommerce merchants, shoppers, or third-party developers.

But that's not the only award this team snagged.

They also won the Sorting Hat Award for the best use of AI across any of the projects in the US and Europe.

Wondering why we came up with these names for our awards? Learn all about these bi-annual events in the Hackathon Details section of part one in this series.

If you're ready to dig into this project, let's begin by meeting the team behind this magical tool.

Meet Bertie Bott’s Bit Bandits

Robert D. Banbanaste, Software Engineer II on the Control Panel Experience Team

Background: I got my start in business and sales and eventually (thankfully) transitioned into engineering. I joined BigCommerce in 2022, with fresh legs and a move from Miami, Florida to Austin, Texas.

Fun fact: The Control Panel Experience Team is an amalgamation of passionate nerds. Also, p-adic numbers.

Role in this project:  I helped build the API. I set up and allocated infrastructure. I designed and styled out the frontend as well.

Connect with Robert on LinkedIn.

Mitch Lombardi, Software Engineer II on the Control Panel Experience Team

Background: I grew up in New England and graduated from the University of Connecticut in 2018. I moved to Austin, Texas when I joined BigCommerce in 2022.

Related Article: Why Mitch Calls Datadog the Web Team's Best Friend

Fun fact: I love finding and learning new hobbies. My current hobby is building custom mechanical keyboards.

Role in this project: I acted as the subject matter expert (SME) for all things AI. I helped build the API and ingestion engine (which embeds all the documents and stores them in a vector database) for the project.

Connect with Mitch on LinkedIn.

Kevin Carr, Software Engineer II on the Control Panel Experience Team

Background: I made a significant career shift in 2019, transitioning from a construction background to my true passion: coding. After successfully completing an 18-month coding bootcamp, I secured my first engineering position with BigCommerce in 2021, marking an exciting milestone in my professional journey.

Related Article: See How Kevin Overcame Adversity

Fun fact: When I was 15 years old, I got invited to participate in a halftime competition show during a Houston Rockets basketball game. Not only did I lose the competition, I also fell flat on my face in the process.

Role in this project: I handled some of the configuration that integrated the new design within our existing onboarding layout. I also coded the spinning loading animation.

More from Kevin: See How He Went from No Code to a Software Engineer

Connect with Kevin on LinkedIn.

Adam Bilsing, Senior Software Engineer on the Control Panel Experience Team

Background: I went to school for biology and started at BigCommerce in 2011 on the support team.

Fun fact: I grow coral as a hobby and live with many birds.

Role in this project:  I created the onboarding experience that is triggered to serve the UI, and I helped build out the frontend.

A.b. Martinez, Senior Software Engineer on the Control Panel Experience Team

Background: I started my career with BigCommerce almost a decade ago as a Customer Support Agent. I left to pursue a career in software engineering and returned a little over a year ago due in large part to the company’s culture, its people, and leadership.

Open Roles: Explore Your Next Engineering Opportunity

Fun fact: I’m currently training to run a half marathon at the end of this year. I do not enjoy running long distances… at all. However, I promised my wife I would run one with her, so here we are. 😃

Role in this project: I helped build the crawler, which collected content from various sources and formatted it so that it could be used to train our model.

Tommy Ekstrand, Software Engineering Manager on the Control Panel Experience Team

Background: I joined BigCommerce in 2012 in Customer Support and now manage our Control Panel Experience Engineering Team.

Fun fact: The difference between a hippo and a zippo is that one is really heavy and the other is a little lighter.

Role in this project: I helped organize everyone into workstreams, built part of the scraper, and built the Algolia connector (not really mentioned in the article, but is just another way to discover content versus the large language model [LLM]).

Connect with Tommy on LinkedIn.

Bertie Bott’s Bit Bandits’ Project

Bertie Bott’s Bit Bandits kept the team dynamic alive by sitting down as a group to answer all of the questions together. As a result, you'll be reading the most concise answers out of the five winners so far — just further validation on how this team continues to think about the user experience. 😉

How did you come up with your team name?

We used ChatGPT, just like the Magical Merchandisers team.

Please describe your project.

It's an AI-assisted knowledge chatbot. We ingested content from BigCommerce content sources into a vector database and used ChatGPT to search results in a conversational form.

Related Article: Hackathon Winners Part 4 - Product Taxonomy in Just One Click

What inspired you to choose this project?

AI is blowing up, and it seemed cool. Originally, we had a different idea, but we were limited by security concerns.

Which pain points does your project solve?

  • Manually scouring through documents.
  • Technical support inside our company.
  • Knowing where to go to find content (blog, support, developer documentation, etc.).
  • Knowing specific BigCommerce lingo/jargon in order to find the correct content.

Which challenges did you face while you were developing your project?

  • Design — We’re a bunch of engineers, after all.
  • Ingestion of content/parsing html scraped from web pages into subsections was tricky.
  • AWS Lambda functions have limited memory.
  • GPT-4 is slow.
  • Ingesting content from many different web sources into a single-data type took a while.

How did you overcome these challenges?

  • A cool loading animation that we found online!
  • Decided the slowness was acceptable for the Hackathon demo.
  • Reduced the number of dependencies and built more in-house.

What’s the one thing you’d change about this project?

  • Better UX and a UI that is more available across that entire Control Panel, rather than just the dashboard.
  • Speed up the search through various means.
  • Better markdown parsing for responses.
  • Add more references to our Hackathon theme, a magical world that rhymes with Scary Otter. Due to legal reasons, you'll have to use your own wand to figure out which franchise we mean.
Still not a Scary Otter

What are the next steps for this project?

  • Get security and legal approval to deploy these new tools.
  • Design, review, and rebuild the UI.
  • Add eventing to understand use.
  • Add feedback tooling similar to OpenAI chat to measure usefulness and help train the model.
  • Create a standalone interface to be accessed by customer-facing BigCommerce employees.

What advice would you give to an engineer who is about to participate in a Hackathon?

  • Move fast and break things (e.g. push directly to main, deploy quick).
  • Give your teammates the freedom to explore what they find exciting. You’re likely to find even better things you had never thought of.
  • Hack in person. The collaboration is immeasurably better.
  • Find ways to split the work into streams that can be combined later.
  • Sit down together and do a technical design up front to identify a broad solution before writing code.

If any engineers would like to contact you, how should they reach out to you?

Reach out on LinkedIn (using the links in our bios) with a message written by ChatGPT. 😉