Back to customer stories

How Yum! Leverages LinearB AI to Save 321 Dev Hours/Month

Yum! Brands, a multi-national fast food corporation, leveraged LinearB workflow automation and AI to automate over 35% of their pull requests and save their developers 321 hours per month.
Results
35.8%
of all PRs automated
30
dev minutes saved per PR
321
dev hours saved/month

Industry

Food Services

Country

United States

Company Size

Enterprise

Introduction

As a Senior Engineering Manager at Yum! Brands' Pizza Hut Digital Ventures (PHDV), Alex L. and his team were deeply involved in the complex migration of key Pizza Hut markets from legacy e-commerce platforms to a modern, composable architecture. This involved building sophisticated front-end experiences designed to integrate with a new, centralized Yum-built backend ordering pipeline, while also handling critical brand-specific, market-level integrations such as POS, payments, and authentication. Around October 2024, the strategic landscape shifted significantly. The wider engineering group Alex was a part of saw its mission expand dramatically, moving beyond Pizza Hut platform consolidation to become a central element of Yum!'s ambitious, company-wide BYTE strategy. The goal evolved: create a unified, scalable, multi-brand front-end platform for Pizza Hut, KFC, and Taco Bell globally. A key technical challenge was ensuring this unified platform could also support significant market-level customization. Meeting the distinct brand identities, localized user experiences, and specific UI requirements of individual markets relied heavily on leveraging a sophisticated, shared design system and templating capabilities developed concurrently by dedicated design and engineering specialists. This strategic pivot, compounded by the need to effectively integrate and utilize these shared components for both global consistency and local flexibility, alongside an organizational restructuring directed by senior product and engineering leadership that formed the new Core platform and Market Activation teams, introduced new layers of complexity. For Alex, this meant guiding his engineers to effectively balance foundational platform development (which included consuming and contributing to the evolving design system and templating frameworks) with ongoing market activation demands, managing intricate dependencies, and maintaining delivery predictability across distributed contributors. To better equip his teams to navigate this dynamic environment and foster data-driven improvements within their workflows, Alex actively sought out tools like LinearB to enhance visibility and streamline processes. To achieve their goals, Yum! implemented a three-step strategy:
  • Step 1: Improve Planning Accuracy with data-driven conversations
  • Step 2: Automate 35.8% of all PRs with workflow automation
  • Step 3: Enhance retrospectives with AI-driven insights

Step 1: Improve Planning Accuracy with data-driven conversations

Before adopting LinearB, Yum! didn’t have a data-backed view into where engineering time was being spent. Meetings dominated their schedules, leaving little room for coding and Alex’s team found themselves working overtime to hit deadlines. In order to improve his team's Developer Experience, Alex wanted to improve predictability across his org. He started by tracking two key metrics: Planning Accuracy and Capacity Accuracy.
  • Planning Accuracy: The ratio of planned work vs. what is actually delivered during a sprint or iteration. High planning accuracy signals a high level of predictability and stable execution.
  • Capacity Accuracy: Measures how many issues (or story points) a team completed in an iteration (planned or unplanned) compared to the amount planned for that iteration.
Improve commitment on planned work: Accuracy Scores over time.
Alex and his team used these metrics as signals to detect potential issues within their development process. By tracking these key indicators, Yum! gained real-time insights into team performance, allowing them to spot inefficiencies and take corrective action before their sprints got derailed. This newfound visibility also enabled them to move beyond anecdotal feedback and instead base discussions on data-driven insights.
“We’ve improved our communication between teams and the way we conduct our sprints. LinearB helps provide a common, data-backed language that my Product Manager, Tech Lead, and I use to improve how we conduct our sprints and planning within the teams I oversee. Having objective metrics readily available facilitates a shared understanding of capacity and progress, helping us define 'what good looks like' for our workflows and reducing time spent purely on alignment in meetings.”
Photo of Alex L.

Alex L.

Senior Engineering Manager, Yum!
This approach also fostered better collaboration between product and engineering teams. By introducing shared KPIs, they ensured that both teams were aligned.
“If I see that planning accuracy numbers are low, I can look through the tickets and delve deeper, which then helps me have a conversation with my tech lead and product manager. These kinds of conversations wouldn’t happen otherwise. But with LinearB, the metrics are right in front of us. It shifts the focus to ‘What can we do differently?’ rather than the issue being forgotten and accumulating over time.”

Step 2: Automate 35.8% of all PRs with workflow automation

As Yum! gained more visibility, they recognized that manual processes were slowing them down. PR reviews were inconsistent and engineers were spending too much time on administrative tasks rather than coding.
“We had to maintain a tight balance between delivering features for existing markets, working on the core of the platform, and fixing bugs. If we have an uptick in bugs, we need to ask: Why? Is it an issue with our development processes? Are we prioritizing the right things?”
By implementing WorkerB (Linearb’s bot assistant) and gitStream (LinearB’s workflow automation engine), Yum! was able to add the following automations into their processes:
  • Code Experts: Ensuring PRs are automatically routed to and reviewed by engineers with the most knowledge on the core platform code.
  • WorkerB Automated Notifications: Alerting engineers when PRs need attention to reduce idle time and tighten review cycles.
  • Estimated Review Time Labels: Adding color-coded labels to PRs with estimated review time, so developers can prioritize their time accordingly.
Bug fix: resolve race condition that affects some user logouts
With these new rules in place, Yum! Was able to automate over 35% of their PRs and reduce their overall cycle time. This allowed their teams to focus on high-value work while minimizing disruptions, all while saving their developers 321 hours per month.

GitStream ROI

Number of PRs merged/month
1793
Gitstream automates 35.8% PRs (Patch, minor, and rubber stamp PRs)
642
Minutes per PR saved (includes context switching waste)
30
Total dev hours automated/month
321

Step 3: Enhance retrospectives with AI-driven insights

One of the most significant transformations was how the team approached retrospectives. Instead of reviewing sprints based on scattered recollections, they used LinearB’s AI-driven summaries to drive discussions rooted in data.
“It’s like a lighthouse beacon. It doesn’t show us something new, but it forces us to pay attention to the right things. You can’t argue with the numbers. If the data behind them is wrong, then at least we’re having those conversations and improving as a result.”
By leveraging LinearB’s AI-powered sprint summaries, the team saw immediate improvements in how they conducted retrospectives and sprint planning.
“Lately we’ve had quite a lot of discussion with the AI-powered Linearb Sprint Summary. We’ve noticed people have been caring about those metrics a lot more than before. I haven’t found anybody who didn’t like the AI summary. I mean, our Director of Product read through the summary and said, ‘This is exactly what I’m thinking. I’m glad the AI agrees with me.’ We started comparing the AI-generated summaries to our actual sprint progress, and it made us look deeper. We’d spot differences and ask: Why is this blocked? Could we have done something differently at the start of the sprint?”
AI Iteration Summary: Improve Planning, Low Capacity Accuracy.
AI-generated retrospectives analyze past sprints and provide automated insights into team performance, like carryover items and major accomplishments. These conversations led to actionable changes, like proactively breaking down features into components to avoid unnecessary blockers. They also improved transparency between engineering and product management, ensuring that stakeholders had a clear, shared understanding of sprint progress.
“Where the data and automation really help communication for us is smoothing out the code review process. You know, we have our Core platform teams and Market activation teams all working in the same big monorepo. Pull requests are really where those different workstreams meet, and things can easily get tangled or delayed. Using the insights and the workflow automations we've set up helps us see where reviews might be stuck, route things smarter, and just get code merged more smoothly between the different groups. It cuts down a lot of friction right at that integration point.”
Photo of Alex. L

Alex. L

Senior Engineering Manager, Yum!

More stories

Read more
syngenta wide
rabbite care wide
super wide