Honeybee uses large language models to make localization tasks more cost efficient

Centific, a world-class digital solutions provider today announced the launch of Honeybee, a toolkit that supports multipurpose AI systems that perform a wide range of tasks such as supporting localization of content. Additionally, Honeybee helps reduce localization costs by performing essential duties such as evaluating machine translation accuracy, making language-based and localization tasks up to 50% more efficient.

As more businesses leverage large language models (LLMs) to train AI using massive amounts of data to understand, evaluate, classify and generate human language, Honeybee finetunes LLMs to make the process faster and more effective. To ensure Honeybee applies LLMs responsibly, Centific uses a human-in-the-loop approach, with people performing crucial roles such as evaluating the model, ensuring human alignment and de-biasing the model.

Marketing Technology News: MarTech Interview with Anudit Vikram, Chief Product Officer at Intentsify

“Honeybee advances AI by giving businesses a useful toolkit for practical applications such as localizing high volumes of complex content,” said Venkat Rangapuram, CEO, Centific. “Multipurpose AI is revolutionizing many industries by performing more complicated tasks, and Centific is already helping clients reap the benefits.”

Tested for several use cases via human-in-the-loop validation, Honeybee is showing strong results in machine translation quality evaluation for specific language combinations. (See chart to the right)

In addition to performing machine translation quality evaluation, Honeybee performs many other tasks. These include:

• Query/text classification
• Personally identifiable information (PII) discovery
• Synthetic data creation
• Content toxicity assessment
• Sentiment analysis
• Data labelling
• Named entity recognition

Centific currently uses Honeybee with Fortune 500 clients on an array of customer experience, data, intelligence, and IT services, continuing to prove the solution across use cases and at scale.

Marketing Technology News: Creating a Personalized Customer Journey