top of page

AI Translation Solution for a Non-Profit Association (2025)

Trained and implemented an AI interpretation tool that improved French attendee accessibility while saving $8K annually in interpreter costs for a big association with 150,000 members.

Impact

  • Discreet yet accurate live interpretation improved user experience.

  • Using AI tool saves 8,000 in yearly costs moving forward.

  • Time required to set up is decreased by 30% each time the tool is trained for a company event.

Problem

An association hosts 2 to 3 in-person events each year for it's members. Because these events serve both English and French-speaking audiences, interpretation plays an important role in making the experience accessible and inclusive.


The previous interpretation setup created several challenges:​​

  1. Live interpretation induced significant costs.

  2. French-speaking attendees had to wear large headsets, which made them stand out in the room, and speaking during sessions created background noise for nearby attendees.

  3. Event terminology required highly specific bilingual context.

Goal

Train Wordly AI, an AI-powered live interpretation solution with a highly specific terminology list that can be reused for future events.

Strategy

Firstly, I needed to learn how to train the AI tool, then work with the Translations team to determine terms needed to be translated, and lastly train and troubleshoot the tool until live interpretation was accurate.

Execution

1. Identified terminology that needed human oversight

AI interpretation works best when it understands the context of the organization. CREA-specific terms, acronyms, names, product references, event names, and regional French preferences could not be left to default AI interpretation.

For example, when speakers said “CREA,” the AI could misinterpret it as “Korea.” These kinds of errors could create confusion for attendees and reduce confidence in the tool.

To prevent this, I worked with the Translations team to identify the terms most likely to affect interpretation quality. 


2. Built and trained a custom terminology database

I created and maintained a terminology database of approximately 4,000 English and French terms.

This included:

  • Company-specific terminology

  • Event names

  • Speaker names

  • Product names

  • Acronyms

  • Real estate industry terms

  • Quebec French preferences

  • Common misinterpretations

  • Approved replacement terms

This training helped Wordly AI better understand the language CREA uses across events and improved the accuracy of the attendee experience. 


3. Created replacement and block lists

One of the most important parts of the project was building replacement rules.

When the AI misunderstood a word or phrase, I added instructions telling the system what to use instead. For example, if the AI heard “Korea” when the speaker meant “CREA,” I created a replacement rule to correct the output.

I also created a block list to prevent inappropriate, incorrect, or unwanted words from appearing in live translations.


4. Tested the output with the Translations team

After training the system, I met with the Translations team to review whether Wordly AI was interpreting key terms correctly.

This feedback loop helped improve the quality of the tool before it was used live. It also made the implementation more reliable because the translation experts were involved in reviewing the output, while I managed the technical setup and AI training process.


5. Supported live event delivery

During events, the Events team connected and used Wordly AI as part of the session setup. I remained available on site to help troubleshoot if any issues came up.

This ensured that the solution was not only prepared in advance, but also supported during real-time use.

Results

  • Saved approximately $8,000 annually in interpreter costs

  • Built a reusable terminology database of approximately 4,000 trained termsReduced future preparation time by allowing the team to build on an existing AI training foundation

  • Improved the French attendee experience by making interpretation more discreet and accessible, reduced background noise and disruption in live session rooms

Skills

  • AI tool training, Time Management, Innovation, Teamwork

  • French

bottom of page