> Markdown version of [/videos/622-useful-ai-friends-for-developers-building-a-multilingual-app](https://www.wearedevelopers.com/videos/622-useful-ai-friends-for-developers-building-a-multilingual-app). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# Useful AI friends for developers – building a multilingual app

Why do general LLMs fail at app localization? Discover how to securely combine Copilot, GPT-4, and DeepL to modernize legacy code without inflating token costs.

- **Speakers:** [Andreas Erben](https://www.wearedevelopers.com/@andreas-erben), [Tim Cadenbach](https://www.wearedevelopers.com/@tim-cadenbach)
- **Event:** World Congress 2023
- **Published:** August 11, 2023
- **Duration:** 30:27
- **URL:** https://www.wearedevelopers.com/videos/622-useful-ai-friends-for-developers-building-a-multilingual-app

## Summary

Andreas Urban and Tim demonstrate how to modernize a legacy application using a combination of contemporary automated tools. Structured around a fictional scenario of inheriting an outdated Windows Forms app, the talk walks through utilizing the ChatGPT code interpreter to safely unpack and inspect untrusted zip files. From there, the session explores applying GitHub Copilot for code analysis before escalating to Azure AI Studio's GPT-4 model. By leveraging system messages for prompt grounding, engineers can successfully guide the AI to restructure existing front-end architectures and rebuild clunky user interfaces without tripping content filters.

Beyond automated refactoring, the presentation explores the nuances of software localization workflows. The speakers demonstrate extracting hardcoded legacy strings into an i18n-compatible JSON format, immediately exposing the pitfalls of using general-purpose large language models for direct translation. Single-word elements like "home" or "resume" carry heavy ambiguity, leading LLMs to hallucinate or generate contextually inconsistent text. Because general models process translations differently on each run, they disrupt reliable automation flows and inflate infrastructure costs through massive token usage.

To overcome these localization barriers, specialized tools offer greater predictability. By integrating a dedicated translation service like DeepL, development teams can inject industry-specific context and construct tailored translation glossaries to handle brand-specific terms or regional dialects. This strategy produces human-grade multilingual apps while streamlining the localization payload. The discussion ultimately advocates for matching specialized APIs with general LLMs based on task requirements, while maintaining strict proprietary code security through enterprise-partitioned AI environments rather than processing sensitive data through free public models.

**Keywords:** AI-assisted refactoring, legacy code modernization, chatgpt code interpreter, azure AI studio, github copilot constraints, UI layout generation, LLM prompt grounding, i18n string extraction, software localization workflows, translation context ambiguity, custom translation glossaries, multilingual app development, LLM token optimization, proprietary code security, enterprise AI environments

## Chapters

1. **Evaluating and sanitizing legacy code artifacts safely** (03:02) — Analyzing unknown project archives automatically prevents malicious code execution in local environments.
1. **Restructuring application interfaces using directed system prompts** (06:03) — Providing strict instructional constraints to generative models ensures accurate programmatic layout adjustments.
1. **Refactoring default UI component names for better readability** (11:03) — Leveraging contextual code completion automates the tedious renaming of autogenerated project variables.
1. **Automating the extraction of static text strings** (13:07) — Migrating hardcoded literal values into structured JSON objects establishes a clean localization foundation.
1. **Mitigating automated translation ambiguities through domain context** (16:04) — Employing specialized translation APIs with custom glossaries eliminates errors caused by isolated wording.
1. **Integrating continuous localization workflows into web frontends** (22:22) — Mapping dictionary keys directly to application components allows seamless generation of multilanguage updates.
1. **Securing proprietary source code when using language models** (27:21) — Utilizing enterprise cloud environments guarantees data privacy policies compared to public generic models.

## Related Moments

- [Leveraging AI tooling for accelerated learning and productivity](https://www.wearedevelopers.com/videos/882-coffee-with-developers-dan-mindru-product-marketing-for-developers) (from "Coffee with Developers - Dan Mindru - Product Marketing for Developers")
- [Migrating legacy applications using generative AI](https://www.wearedevelopers.com/videos/1605-navigating-application-modernization-leveraging-gen-ai) (from "Navigating Application Modernization - Leveraging Gen-AI")
- [Using AI copilots to explain and debug legacy codebases](https://www.wearedevelopers.com/videos/1302-wearedevelopers-live-dishonest-charts-britcss-debugging-with-ai) (from "WeAreDevelopers LIVE - Dishonest Charts, BritCSS, Debugging with AI")
- [Leveraging artificial intelligence to accelerate system migration](https://www.wearedevelopers.com/videos/1444-crew-management-system-for-airlines-plan-duties-for-pilots-flight-attendants-worldwide) (from "Crew Management System for Airlines: Plan duties for pilots & flight attendants worldwide")
- [Analyzing cloud-based AI code completion architectures](https://www.wearedevelopers.com/videos/961-beyond-autocomplete-local-ai-code-completion-demystified) (from "Beyond Autocomplete: Local AI Code Completion Demystified")
- [Leveraging large language models for code optimization and development](https://www.wearedevelopers.com/videos/1106-the-future-of-computing-ai-technologies-in-the-exascale-era) (from "The Future of Computing: AI Technologies in the Exascale Era")

## Related Articles

- [MLOps And AI Driven Development](https://www.wearedevelopers.com/magazine/82-mlops-and-ai-driven-development)
- [Exploring AI: Opportunities and Risks for Developers](https://www.wearedevelopers.com/magazine/522-exploring-ai-opportunities-and-risks-for-developers)
- [How we Build The Software of Tomorrow](https://www.wearedevelopers.com/magazine/120-how-we-build-the-software-of-tomorrow)
- [Dev Digest 132 - Binging WADFlix?](https://www.wearedevelopers.com/magazine/473-dev-digest-132-binging-wadflix)

## Related Jobs

- [AI Software Engineer (Germany)](https://www.wearedevelopers.com/jobs/48317-ai-software-engineer-germany) at **Sunhat**
- [AI & Machine Learning Engineer (all genders)](https://www.wearedevelopers.com/jobs/48217-ai-machine-learning-engineer-all-genders) at **msg**
- [Staff Software Engineer, Copilot Experiences](https://www.wearedevelopers.com/jobs/ext/164361-staff-software-engineer-copilot-experiences) at **GitHub**
- [Principal Software Engineer, Enterprise AI Platform](https://www.wearedevelopers.com/jobs/ext/1467292-principal-software-engineer-enterprise-ai-platform) at **GitHub**
- [Senior AI Agent Software Engineer (Go, Python) (m/f/x)](https://www.wearedevelopers.com/jobs/48277-senior-ai-agent-software-engineer-go-python-m-f-x) at **Dynatrace**
- [AI Operations Manager (all genders)](https://www.wearedevelopers.com/jobs/48263-ai-operations-manager-all-genders) at **envelio**