Last Tuesday, senior localization leaders, AI experts, and global content strategists from leading multinational companies gathered in Shenzhen for the first Chinese edition of Loc360°. Hosted by Phrase and Clearly Local, the event explored AI’s disruptive impact on our industry. From high-level keynotes on the future of localization and human linguists to expert panel discussions on practical AI use cases and deployment strategies, each session provided a rare opportunity for participants to openly discuss the industry’s direction and leave with smarter strategies for leveraging AI.
If you missed the event or just want a recap, we’ve got you covered. Here’s our ten biggest takeaways.
1. AI Is Transforming Localization from a Post-Production Step to a Core Strategy
In his keynote speech, Phrase CEO Georg Ell emphasized that AI is shifting localization from being an afterthought in content creation to a central driver of global strategy. No longer confined to the end of the pipeline, AI now enables hyper-personalized content—delivering the right message in the right language, channel, and cultural context—to boost conversion rates and market penetration.
2. Human-AI Collaboration Remains Non-Negotiable for Quality
Clearly Local’s Managing Director, Philippe Cao, said that while AI excels at automating repetitive tasks, after more than four years of experimenting with and developing various AI solutions, none have been able to fully replicate human-expert quality. He emphasized that human expertise remains critical for quality control and cultural nuance. To illustrate this point, he highlighted the case of Swedish fintech company Klarna, which recently reversed its decision to replace 700 employees with AI due to severe quality issues.
3. The Industry Is Divided Between Conservative and Expansionist Approaches to AI
Philippe explained that companies are adopting AI in two primary ways:
- Conservative: Doing more with fewer resources (e.g., maintaining content volume with smaller teams).
- Expansionist: Scaling content exponentially with the same team (e.g., hyper-personalized marketing).
Most fall somewhere in between, but competitive pressures are pushing trends toward expansion and hyper-personalization.
4. Technical Challenges in AI Adoption: Quality, Context, and Cost
Key hurdles include:
- Inconsistent Quality: AI may produce logical errors or ignore style guides, requiring extensive fine-tuning.
- Contextual Limitations: AI struggles with complex scenarios like cross-page UI interactions or idiomatic language (e.g., mistranslating regional humor).
- Cost of Tuning: Initial productivity gains from AI can be offset by the time spent optimizing models, especially with incomplete testing.
5. Phrase’s Ecosystem Vision: A Scalable, Integrated Localization Platform
For AI to be effective and used at scale, it needs to meet users where they are. Georg mentioned how the idea of the traditional TMS is now outdated, and modern TMS like Phrase are quickly integrating AI, APIs, and third-party tools like Amazon Market Place to automate content translation for global brands. With quarterly product launches and a focus on hyper-automation, Phrase aims to be the “Microsoft of language technology,” according to Georg, empowering clients to build customized workflows with features like Auto Adapt while tapping into expansive partner ecosystems.
An AWS panelist highlighted the symbiotic relationship between Phrase and AWS, commenting that the collaboration strengthens both ecosystems. They noted that AWS’s customers need localization solutions to:
- Comply with local laws and regulations for market entry,
- Support minority languages, and
- Address industry-specific requirements.
As a cloud provider, the panelist explained, AWS can’t deliver all these specialized capabilities alone. By integrating third-party services like Phrase into their platform, they help customers focus on their core business while ensuring professional localization across verticals.
6. The Rise of the GCSP (Global Content Solution Provider)
Companies like Clearly Local are evolving from traditional LSPs to global content solution providers, offering end-to-end services beyond translation such as video production and UI/UX testing. This shift is driven by client demands for integrated solutions that reduce workflow handoffs between teams.
7. Interactive Polls Revealed Audience Pain Points
A live poll at the event showed nearly all attendees identified the following as major pain points:
- Localization Efficiency: Most participants identified the poor quality of source texts as the most wasteful area in the localization process, as they lead to excessive rework. The second most cited issue was the lack of terminology or style standards, resulting in inconsistent content. Additionally, long-neglected translation memory optimization was noted for degrading pre-translation quality and proofreading efficiency.
- Cross-Departmental Cooperation: Overseas distribution and localization team members reported that the biggest obstacle to cross-departmental cooperation is internal underestimation of localization’s importance, leading to insufficient support for standardized processes and limited influence in product development decisions. The second largest challenge is operational inefficiencies, including absence of automation tools and disjointed systems with unsynchronized data and incompatible interfaces.
- Product Localization: Product team members identified two major challenges in localization: lack of visual context for UI translation and localization bottlenecks impacting product release schedules.
- Procurement: Procurement department members see the fundamental challenge in sourcing localization platforms and services as the persistent difficulty in quantifying return on investment (ROI). This measurement problem becomes particularly acute given two competing pressures: the organization’s rising expectations for AI-enhanced localization quality and its simultaneous mandate for annual cost reductions.
- AI Improvements: All participants agreed that AI-powered quality assessment tools have most significantly improved their workflow by enabling rapid error detection and modification.
8. The Future: AI Agents, Cross-Dimensional QA, and Role Evolution
Speakers predicted that AI will evolve into autonomous agents capable of planning complex tasks (e.g., breaking down localization workflows into steps). Additionally, tools like Phrase’s AutoLQA will expand beyond text to scan webpages and images, enhancing cross-dimensional quality assurance. For professionals, roles will shift from “executors” to “technology integrators,” focusing on prompt engineering and user experience.
9. Common Missteps in AI Adoption: Overestimating Autonomy, Underinvesting in Data Quality
Speakers warned against:
- Treating AI as a “set-it-and-forget-it” solution (e.g., assuming generic MT models will meet niche industry needs without customization).
- Neglecting data hygiene: AI outputs are only as good as input data (e.g., outdated TMs or incomplete style guides lead to inconsistent translations).
Solution: Invest in iterative testing, collaborate with LSPs to refine AI models, and prioritize contextual data in training.
10. AI Powers End-to-End Localization
During the AI workshop with brands like OPPO, Xiaomi, Segway and Sungrow, participants highlighted how AI is now embedded in every stage of localization—from terminology extraction to final quality checks. Here’s what the experts shared:
- Terminology: AI is used to extract terminology from content, but short terms are more prone to polysemy (e.g., a single word with multiple meanings in specialized contexts). Workshop attendees emphasized that while tools like Phrase’s AI term extraction assist in identifying potential terms, human native speakers with industry expertise must review and edit them for contextual accuracy.
- Translation: Historical methods like OCR and manual patching for static screenshots are being replaced with AI-driven workflows for efficiency.
- Quality Assurance: AI-powered LQA tools that scan translations for errors like inconsistent terminology or brand names are playing a greater role in ensuring quality. Phrase’s Auto Adapt feature (launched this year) was demonstrated to refine tone via prompts, reducing post-editing effort.
- Workflow Automation: Phrase’s Orchestrator was highlighted as a powerful platform for building customized localization workflows. While it features a visual interface that helps simplify orchestration, attendees noted that effective implementation often benefits from collaboration with technical teams. The result is greater automation, smoother handoffs, and improved control across multilingual processes.
Conclusion: Balance Innovation and Humanity in Localization
Loc360° Shenzhen reinforced that AI is a catalyst, not a replacement, for human expertise. As Clearly Local and partners like Phrase demonstrate, the key to success lies in balancing AI-driven efficiency with human creativity and cultural insight. By embracing collaborative ecosystems, iterative testing, and a people-first approach, the industry can unlock the full potential of global content in the AI era.
