Case story: Integrating Azure AI with Microsoft Teams
A story of how we integrated Azure AI Custom Translator into Microsoft Teams
Translations: a common challenge in projects
When working as a consultant across diverse customer domains, one recurring issue I’ve encountered is translations—whether for app UIs, reports, or internal communications. Traditionally, translations are handled manually, often inconsistently, and can easily become a bottleneck in implementation work. After completing my Azure AI Engineer certification, I started to think: Could we leverage AI-powered translation to streamline this process?
Starting with a simple proof of concept
The initial idea was simple: create an Azure AI Custom Translator with a small amount of baseline translations to handle subsequent translation changes and additions. I built a proof of concept (PoC) using Python and the Azure AI Translator API. This simple tool worked well for short UI translations, significantly reducing the manual effort of SMEs and developers.
Scaling up the AI solution to meet customised business needs
After seeing the translator tool in use, the customer asked if it could be expanded for their internal translation needs. They had a constant demand for translations and wanted something more reliable and scalable than ad hoc freelance work.
To make the AI-powered translator truly useful, we needed to train a custom AI translation model using Azure pre-trained model for their domain. Training requires at least 10,000 sentence pairs (about 500 pages) of domain-specific material. A mix of hard work, co-operation and some further AI assistance helped us to train this more sophisticated translation model.
Integrating the AI solution seamlessly into Microsoft Teams
Once we had a functional translation model, the next question was how to make it easily testable and accessible. The answer was Microsoft Teams. We used Teams Toolkit to build a Teams bot app that could:
- Translate text on demand via messaging
- Translate whole files via attachments
- Integrate seamlessly into the customer’s workflow
Deploying the AI translator as a Teams bot would allow users to access it directly within their daily tools, making AI assistance more accessible and intuitive.
Driving real business value with AI
This project highlights how small, targeted AI helper tools can provide business value. By leveraging Azure AI and Microsoft Teams, companies can create powerful, practical AI solutions without massive development efforts. Additional tools could include:
- Azure Document Intelligence combined with a Microsoft Teams bot
- Automating the reading and processing of standardised forms (such as invoices, forms, or receipts), streamlining workflows and reducing manual work and errors.
- Azure AI Search and OpenAI embedded directly into Microsoft Teams
- Helping employees find quick, accurate answers by searching internal company knowledge directly from Teams with the tools that they already use makes work more efficient and removes the hassle of manually digging through internal company knowledge and documentation.
AI is not just about big, complex models but also small, practical tools that solve real-world problems. The potential is huge, whether it's automated AI translations, handling standard forms, AI-driven knowledge retrieval, or something else. AI can become a natural part of everyday business operations with the right approach.
Are you ready to explore AI-powered productivity tools in your organisation? Let’s talk!
Share on:
Want to read more about how Rebase gives back? Check out our other stories
Mastering Authentication Flow
"Tired of frequent logouts disrupting your app experience?"
Mastering Authentication Flow
"Tired of frequent logouts disrupting your app experience?"
