Back to blog
AI·November 3, 2020

How I Built a Self-Learning Agile Slackbot to Help Allegiant Air's Teams Embrace Scrum

When Allegiant Air transitioned from waterfall to Agile, the hardest part wasn't the process — it was getting everyone speaking the same language. So I built them a patient, always-on teammate powered by IBM Watson.

The Real Barrier to Agile Adoption

When Allegiant Air decided to move from waterfall to Agile, leadership assumed the technical challenges would be the hard part. They weren't. The harder problem was cultural and linguistic.

Teams across the organization were suddenly being asked to participate in sprints, estimate story points, run retrospectives, write acceptance criteria — and most of them had no shared vocabulary for any of it. Agile terminology without context is just jargon. And jargon creates friction, confusion, and quiet disengagement.

Slide decks and training sessions helped briefly, then faded. The questions kept coming.

A Different Approach

Rather than another knowledge base that nobody would visit, I proposed building something that lived where the team already was: Slack.

The idea was a conversational bot that anyone could ask about Agile — in plain, natural language — and get a clear, contextual answer immediately. No searching, no ticket, no waiting for the Scrum Master to be available. Just ask and get an answer.

How It Worked

The bot ran on Node.js with IBM Watson handling natural language understanding. When a message came in, Watson parsed intent and the system first checked an internal database of previously answered questions. If a match existed, it returned the stored answer instantly.

If no match was found, a Selenium and Java scraper would hit scrum.org in real time, extract the most relevant content for the query, and return a clean, formatted answer directly inside Slack — all within seconds.

The part I was most proud of was the feedback loop. If the user indicated the answer was helpful, the system extracted keywords from both the question and the response and persisted them to the database. Every positive interaction made the next one faster and more accurate. The bot got smarter the more it was used — without any manual curation.

The Result

Teams that had been glossing over Agile terms in standups started asking better questions. Sprint planning got sharper. Retrospectives became more substantive. And because the bot was always available — at 9pm when someone was prepping for tomorrow's planning session, or at 7am before standup — it met people in their moment of confusion rather than asking them to find a resource later.

The Agile transformation had a knowledgeable, patient, always-on teammate. That turned out to be exactly what was missing.

Marcus

Marcus Bass

AI · Career Q&A

get in touch