From zero to global in 6 weeks: How a fitness platform built an AI-driven localization program
“We went from English-only to a global-ready localization program in under six weeks. Intento helped us set up AI-first, requirements-based automation fast, so our linguists could focus on governance and brand fit instead of reworking raw output.”
— Eduardo D’Antonio, Director of Localization, Strava
Fitness technology / Social platform for athletes
Strava had six weeks to build a localization program and ship 2M+ words across 7 languages. They started from an English-only app with no internationalization (i18n), no translation assets (no TMs), and no MT training data or workflows
Crowdin (TMS)
Intento built the translation workflow from scratch in just two weeks, using AI Agents to apply glossary, tone, and formatting rules, ensuring translations met quality requirements without training data and kept the project on schedule—while enabling parallel work across teams to avoid sequential handoffs.
Why Strava needed a global-ready localization program
A fitness technology company (Strava) acquired a UK-based running app, Runna, in spring 2025. At the time, it was English-only, and leadership moved to prepare it for additional markets without delay.
In technical terms, it lacked i18n. That means the code wasn’t set up to easily handle things like multiple languages, date formats, or character sets. For a business, not having i18n means every new market requires extra engineering effort, slowing down time-to-market.
The mandate was clear and time-bound, prioritizing speed over perfection:
- Build a full localization program from scratch
- Choose and implement a modern technology stack
- Onboard a global LSP
- Translate over 2 million words into 7 languages
- All in under 6 weeks
Outcome beyond launch: Within three months of launch, Strava saw meaningful revenue growth and an increase in active users.
Building a localization tech stack in record time
Strava partnered with three providers to move at speed: Cornelius Communications for strategy and orchestration, Crowdin for TMS, and Intento for AI-driven translation workflows.
- Cornelius Communications guided the selection process, helped migrate glossaries and translation memories, and coordinated engineering, product, and localization teams to keep parallel i18n and l10n efforts on track (replacing sequential handoffs).
- Crowdin integrated with 18 content repositories, migrated existing glossaries and translation memories, and set up workflows.
- Intento configured customized AI translation workflows in just two weeks. Without the option to fine-tune models on training data, Intento used its AI Agents to apply glossary rules, tone of voice, and formatting requirements directly into the MT process. This allowed translations to meet business requirements from the start, rather than relying on post-editing alone.
Principle applied: Pragmatism over perfection—launch fast with quality guardrails, then iterate with human review where it matters.
By combining these strengths, Strava could run internationalization and localization in parallel, translating over 2 million words into 7 languages while engineering teams fixed i18n issues in the codebase.
What was achieved in just 6 weeks
- Fully operational Localization Department in 6 weeks.
- 18 repositories integrated into the TMS.
- 2 million words localized into 7 languages.
- Localization and internationalization done in parallel, saving months of delays.
- Teams from both companies (Strava & Runna) trained in i18n and localization workflows.
- Sequential steps were replaced by a parallel, cross-team model to maintain velocity and meet the deadline.
How Strava used Intento Language Hub to enable AI-driven workflows
“Strava had six weeks to ship over 2M words across seven languages. We built the workflow in two weeks, using AI Agents to enforce glossary, tone, and formatting rules directly in the MT workflow—so their translations met their business and language requirements from day one.”
— Konstantin Savenkov, CEO & Co-founder, Intento
- End-to-end flow that prevents rework: Crowdin centralized content, and Intento ensured translation quality through AI-driven customization. Each part connected smoothly, avoiding delays.
- AI that adapts quickly (no training data required): The Intento AI Agents enforced brand terminology, style guides, and tone of voice without requiring months of data preparation, keeping the project on track despite tight deadlines.
- Right role for humans: Kept humans focused on language governance, brand fit, and final decisions—not minor fixes.
- Real-time progress: The Intento workflows produced outputs aligned with the requirements, so reviewers focused on refining rather than reworking, which accelerated delivery.
- Leadership alignment: With strong support from executives, the team could prioritize localization and make fast decisions.
What AI-driven localization made possible in just 6 weeks
This project shows that with the right setup, enterprises can launch a global localization program in weeks, not months.
The company localized 2 million words into seven languages, launched a new Localization Department, and proved that AI-driven workflows can deliver enterprise-grade localization even under extreme timelines—without relying on training data.
Within 90 days of launch, Strava saw significant gains in revenue and active users—validating a speed-first approach and a repeatable model for rapid market entry without legacy assets.