LariBat - Adobe

TDCX and Supa team up on data labelling

TDCX has teamed up with data labelling firm Supa to speed up data labelling with up to 98% accuracy using the human-in-the-loop approach

Singapore’s TDCX has tied up with Supa, a generative artificial intelligence (GenAI)-powered data labelling company, to help organisations overcome one of the biggest challenges in their AI journey – labelling raw data to make it understandable for machine learning algorithms.

According to research by McKinsey, 72% of organisations noted that data management was one of the top challenges that prevented them from scaling AI use cases, while eight in 10 said training AI with data had been more difficult than expected.

Leveraging Supa’s technology for handling large datasets, TDCX said human annotators can reduce data processing time by up to five times, enabling organisations to train their AI models more effectively and generate greater value for their business.

Lianne Dehaye, senior director of TDCX AI, said without accurate, structured and reliable data, organisations will not be ready to leverage GenAI.

With the strong interest in GenAI, many companies find themselves in a rush to benefit from it. However, the truth is, many companies either do not take the critical first step of data labelling or underestimate the resources needed to get it done well. This leads to situations where AI projects end up failing and there is little return on that investment.

“The issue of quality data is even more pertinent in CX [customer experience] applications. Beyond being accurate, there is also a need to ensure that the data is free from bias and takes cultural nuances into account. This is where human intelligence and understanding come in. Our collaboration with Supa strengthens our offerings and will enable us to help clients integrate AI into the CX strategies more quickly and easily.”

Mark Koh, Supa’s CEO and co-founder, said the company’s platform can curate and process large training datasets with up to 98% accuracy for labelled data.

“Achieved through our multi-stage human-in-the-loop approach, this proactive validation process empowers annotators to act as data model teachers, thus minimising potential errors or routing issues,” he added.

Besides supporting the data labelling needs of organisations in a range of industries, including retail, manufacturing and healthcare, Supa’s platform can also work with different data modalities, including images, videos, multilingual text and audio.

Data handled by the TDCX and Supa teams will be managed securely, with organisations retaining all data within their own cloud storage. The two companies are offering a complimentary diagnostic session for companies to understand the opportunities and gaps in their data labelling needs.

Read more about AI in APAC

  • DBS Bank’s AI Industrialisation Programme has been instrumental in industrialising the use of data and AI across its business, resulting in over S$370m of incremental economic benefits.
  • Alibaba’s SeaLLMs are built to address the linguistic diversity and nuances in Southeast Asia, enabling businesses to deploy localised chatbots and translation applications.
  • Malaysian startup Aerodyne is running its drone platform on AWS to expand its footprint globally and support a variety of use cases, from agriculture seeding to cellular tower maintenance.
  • The Australian government is experimenting with AI use cases in a safe environment while it figures out ways to harness the technology to benefit citizens and businesses.

Read more on Data quality management and governance

Data Center
Data Management