Highlights:
- Encord states that its software covers the four critical steps in creating AI training datasets: data management, data curation, model evaluation, and annotation.
- While automated data annotation systems aren’t new, traditional ones depend on human oversight. Encord, however, uses AI to fully automate the process, enabling teams to prepare large datasets for AI training much faster.
Recently, a data annotation platform provider, officially known as Cord Technologies Inc., Encord, has secured USD 30 million in Series B funding. The investment round was led by Next47, with additional support from previous investors Y Combinator, CRV, and Crane Venture Partners.
The startup has developed what it refers to as a “data development platform,” aimed at simplifying the training process for artificial intelligence models. It emphasizes that the success of any AI model is directly dependent on the quality of the data it was trained on, making data preparation and annotation essential tasks for AI developers.
The problem lies in the fact that teams have to manage vast amounts of siloed, uncurated, and unformatted information. Traditionally, data preparation and annotation have been handled manually, creating a significant bottleneck in AI development.
While automated data annotation systems have been around for some time, traditional models still rely heavily on human oversight. Encord takes a different approach by automating the entire process, using AI to oversee the task. This allows teams to prepare large datasets for AI training much faster than previously possible.
Encord asserts that its software simplifies the four crucial steps in developing AI training datasets: data management, data curation, model evaluation, and annotation. By integrating these tasks into a single platform it provides a traceable process that allows developers to analyze their AI models and understand the reasons behind specific outputs. This insight enables them to refine and retrain their models for improved results.
The startup is aiming to compete in the data annotation and labeling industry, which is projected to exceed USD 3.6 billion annually by 2027. However, it faces significant competition. Its main competitor is likely Scale AI Inc., but numerous other startups are vying for a share of the market, including Datasaur Inc., Dataloop Inc., and Soda Data NV.
Eric Landau, Co-founder and CEO of Encord, shared with a prominent media outlet that his company’s platform stands out for its versatility compared to competitors. He highlighted that the platform’s key advantage is its unique capability to help AI developers explore and visualize their training datasets, whether they consist of images, text, video, or voice recordings.
The platform also offers tools to compare the output and performance of different models trained on the same datasets and to detect accuracy issues. Additionally, it can recommend specific types of additional training data to address any problems it identifies.
“Encord lets you consolidate all your data workflows in one platform. Through this consolidation, companies gain traceability that sheds light on the often opaque ‘black box’ of AI, helping to understand why a model makes specific decisions,” Landau explained.
Naturally, Encord is confident that its data development platform is the best, but the positive reception of its product is hard to ignore. The company has already signed up over 120 customers, including notable organizations like Synthesia Ltd., Koninklijke Philips N.V., Zoopla Ltd., Cedars-Sinai Medical Center, and Northwell Health LLC.
He also added, “Encord’s forward-thinking platform addresses one of the biggest challenges in AI today – understanding and managing the data that will give enterprises high quality, reliable outcomes from their AI applications, therefore lowering the risk of AI implementation.”