Pit Stop Predictor

Predict SLA breaches before they happen. Like F1 pit crews timing the perfect stop, our AI analyzes tickets to show which need attention now. Stop reacting, start predicting.

Comment

Pit Stop Predictor

🏎️ What it does

Pit Stop Predictor brings the precision of Formula 1 pit stop strategy to IT service management. Just as F1 teams use telemetry data to predict exactly when a driver needs to pit before tires fail, our app predicts which support tickets will breach their SLA targets before it happens.

Key Features:

  • 🤖 AI-Powered Breach Prediction — Intelligent analysis combining time pressure, ticket complexity, agent workload, and historical patterns
  • 🎯 Speedometer Dashboard — Real-time gauge showing 0-100% breach likelihood, styled like F1 telemetry
  • ⏱️ Pit Window Countdown — Shows the optimal intervention window (e.g., "Pit Window: 2h 15m")
  • 📊 Risk Factor Breakdown — Transparent AI reasoning showing exactly why a ticket is at risk
  • 💡 Actionable Recommendations — Context-aware guidance like "Immediate pit stop required!"
  • 🗂️ Project Dashboard — Bird's-eye view of all tickets sorted by breach probability
  • 🏆 Pit Crew Champions — Podium-style leaderboard celebrating top-performing agents
  • 💬 Rovo AI Agent — Natural language interface: "Which tickets are at risk?"
  • ⚙️ Admin Settings — Customize risk thresholds, factor weights, and themes

🎮 Try It Instantly — Demo Mode

No complex setup required! Every module includes a Demo Mode toggle that generates realistic sample data — tickets at various risk levels, SLA countdowns, agent workloads, and leaderboard rankings.

Perfect for judges, evaluators, or anyone who wants to experience the full app without configuring a JSM project with tickets and SLAs. Just toggle it on and explore!

💡 Inspiration

We've been building apps for Jira Service Management for years, with several solutions already on the Atlassian Marketplace. Through this experience, we've seen a recurring pain point: teams react to SLA breaches after they happen, scrambling to explain missed targets instead of preventing them.

The data to predict breaches exists—time remaining, ticket complexity, agent workload, historical patterns—but it's scattered across screens and requires mental math to interpret.

Then we saw the Codegeist Williams Racing Edition theme.

Every Formula 1 race is won or lost in the pit lane. Teams don't wait for tires to fail—they use predictive telemetry to call the pit stop at exactly the right moment. A few seconds too late means losing positions. A blown tire means disaster.

The connection was immediate: What if support teams had the same predictive power as an F1 pit wall?

Instead of a dashboard showing "SLA Breached" in red, imagine a speedometer showing "78% breach probability—pit window closes in 2 hours." Instead of post-incident reports, imagine knowing which tickets need attention right now.

🛠️ How we built it

Platform: Atlassian Forge with Custom UI

We built Pit Stop Predictor on Atlassian Forge, using the Custom UI approach for maximum flexibility. This gave us full control over the F1-themed experience while leveraging Forge's secure, serverless infrastructure. Pit Stop Predictor is eligible for Runs on Atlassian.

The Prediction Engine

Our core algorithm calculates breach probability using four weighted factors:

Factor Weight Description
⏰ Time Pressure 40% Remaining time vs. SLA deadline
📝 Ticket Complexity 30% Description length, attachments, comments, priority
👤 Agent Workload 20% Open tickets and at-risk items assigned
📈 Historical Patterns 10% Breach rate on similar tickets (past 30 days)

Each factor produces a 0-100 score, combined into a final breach probability. Weights are fully configurable in Admin Settings.

🤖 AI Integration — Rovo Agent

We integrated with Atlassian's Rovo AI for a conversational interface:

  • "Which tickets are at risk of breaching SLA?"
  • "What's the breach risk for SUPPORT-123?"
  • "Show me the queue summary"
  • "Who are the pit crew champions this month?"

🔌 Integrations

  • JSM SLA API — Real-time SLA cycle data
  • Jira Search API — Ticket analysis and historical patterns
  • Forge Storage — Caching predictions (30-minute TTL)
  • Rovo AI Agent — Four custom actions for natural language queries

🚧 Challenges we ran into

  1. JSM SLA API Complexity — Parsing "remaining time" reliably across paused cycles, business hours, and multiple SLA targets required careful defensive coding.

  2. Prediction Accuracy vs. Performance — Multiple API calls per prediction. We implemented a 30-minute caching layer with Forge Storage to balance freshness with performance.

  3. Rovo Agent Integration — Defining action schemas and formatting conversational responses took experimentation to get right.

  4. Demo Mode — Testing without real breaching tickets is tricky. We built a demo mode toggle that generates realistic mock data across all risk levels.

🏆 Accomplishments we're proud of

  1. Complete F1 Design System — We built F1UIKit, a full component library with speedometer gauges, animated needles, podium displays, and 20+ themed icons. The UI genuinely feels like an F1 pit wall dashboard.

  2. Transparent AI Reasoning — No black box. Users see exactly why a ticket is at risk: time pressure at 85%, agent has 12 open tickets, similar tickets breached 40% last month.

  3. Pit Crew Champions — The podium leaderboard gamifies performance, shifting the narrative from "avoiding breaches" to "winning the race."

  4. Natural Language Access — Managers can ask "Which tickets need attention?" and get instant answers without opening dashboards.

📚 What we learned

  1. SLA Data is Messy — Paused timers, business hours, multiple SLAs per ticket, already-breached states. Never assume clean data.

  2. Weighted Algorithms Need Tuning — Time pressure dominates real-world breach risk more than we assumed. Making weights configurable lets teams find their balance.

  3. Gamification Changes Behavior — The Pit Crew Champions leaderboard shifted the app's feel from "breach prevention tool" to "performance platform."

🔮 What's next

  1. 📣 Proactive Notifications — Slack alerts, email digests, push notifications for critical risk spikes

  2. ⚙️ SLA Configurations — Choose which SLAs to track with different thresholds for each

  3. 🎨 Professional Theme — A polished, enterprise-ready aesthetic for teams preferring a cleaner look

  4. 💬 Expanded Rovo — "Reassign at-risk tickets," "Compare this week to last week," "What's causing the most breaches?"

  5. 🏢 Multi-Project Views — Unified risk view across projects for enterprise teams

Built With

Share this project:

Updates