Speak scales its weekly app releases with Tramline

AI-powered language learning app that helps you practice real conversations through daily speaking exercises.

Context

Speak is a language learning app that helps users build confidence through AI-powered conversational practice. By combining speech recognition with real-time feedback, Speak enables learners to improve their speaking skills in everyday scenarios.

The app supports learners in over 20 countries and has crossed more than 10 million downloads. Its presence is particularly strong in South Korea and Japan, where Speak has consistently ranked among the top education apps. Backed by leading global investors, the company is now positioned as a major player in the language learning space. 

With this growth came new demands on infrastructure and internal processes—including a need for faster, safer, and more scalable app releases. The team ships updates to the app every week, which means the release process needs to be stable and easy to understand, not just for mobile engineers, but also for QA, product, marketing, and every other team that cares about what’s being shipped to customers.

Challenges

Before Tramline, Speak managed releases using Git Flow (both the branching model and its accompanying CLI) alongside build pipelines on Bitrise.

While Git Flow is a familiar pattern in software engineering, its CLI came with several limitations. Releases had to be initiated locally by the release manager, who used the CLI for various tasks like creating the release branch, merging it, and tagging production commits. This process bypassed pull requests, leaving little visibility into what had actually changed in a release. The main branch also had to remain unprotected to allow these changes to be pushed: an inherently risky setup, even in a high-trust team.

Our release process worked, but only with a lot of hands-on effort. It became clear we needed a more automated, reliable system that didn’t rely so heavily on individual diligence.
– Sean Kim, Engineering Manager 

Onboarding new engineers onto this manual setup was another challenge. Most team members hadn’t used the Git Flow CLI before, so they had to be trained on the commands and set up the tooling locally.

As the product and team scaled, it became increasingly clear that this developer-centric release model didn’t meet the needs of the broader organization. QA and product teams relied heavily on engineers for updates on what had shipped and how a release was progressing. Granting and managing Play Console access became a time-consuming and error-prone task.

If regular release cycles weren’t already complicated, patchfixes and hotfixes added another layer of complexity. For patches, the release manager had to remember to increment the version correctly, close the current release, open a new one, and manually restart the staged rollout at the same percentage. Hotfixes were just as tedious. Engineers needed to track down the correct commit and branch from the previous release, bump version numbers, generate a new build, close the old release, and create a new one with the updated binary. These one-off processes introduced additional room for error and added stress to an already overloaded release cycle.

There were just a lot of small, manual steps involved in each release. It was easy to overlook something, and hard to feel confident without double-checking every part of the process.
– Sean Kim, Engineering Manager 

The team needed a more streamlined and mature approach to mobile releases: a system that could automate the grunt work, reduce the operational overhead, and give everyone—not just developers—access to the information they needed to stay in sync.

Solutions

Speak adopted Tramline to modernize and streamline its Android release process. Right away, automating the release tasks freed engineers to focus on building the product. Tramline’s workflow engine now handles the entire lifecycle: version bumps, release branch creation, backmerging bug fix commits, and tagging final builds. No more manual Git Flow commands. No more local-only processes.

Questions that once required an engineer are now answered instantly via the Tramline dashboard: What’s the latest internal build? How far along is the production rollout? How does rollout progress map to adoption of the new version? With this level of visibility, most team members no longer need access to the Play Console at all.

With Tramline, releases feel a lot more predictable. We’re not juggling scripts or waiting on someone to run the right commands—everything just flows. It’s taken a huge amount of overhead off the team and made the whole process more transparent.
– Sean Kim, Engineering Manager

Patchfixes are now managed end to end: Engineers simply commit the fix to the release branch, and Tramline creates a new build with the right information and updates the Play Console. Hotfixes follow a similarly streamlined flow. Because Tramline is already connected to the build pipelines, versioning data, and release metadata, hotfixes behave more like lightweight continuations of a previously completed release. This consistency makes hotfixes far less stressful to manage.

Cascading rollouts—a strategy commonly used by Android teams, and something Speak had been handling manually—are also managed by Tramline. Releases only go to 100% once the next version is ready for production, giving the team control to pause rollouts until the last possible moment.

Outcomes

Since adopting Tramline, Speak has transformed its release process from a developer-owned task into a transparent, team-wide workflow. Engineers now spend less time maintaining custom scripts or training others on legacy tools. Instead, the release process is automated where appropriate, with thoughtful guardrails where human input is needed.

At the same time, Speak’s rapid growth and unique internal workflows continue to push the boundaries of what modern release tooling can support. Tramline is working closely with the team to navigate these edge cases, helping evolve the platform and better serve the broader mobile ecosystem.

The move to an automated, team-wide release process has freed up valuable engineering time and reduced friction for QA, product, and other stakeholders. With reliable automation and improved visibility, Speak is better equipped to ship updates to its users, week after week.