Hyderabad-based Founder’s CTGT Raises $7.2M to Help Enterprises Scale AI Beyond Deep Learning

CTGT, an AI startup backed by Y Combinator (YC F24), has secured $7.2 million in funding to drive its mission of scaling AI beyond traditional deep learning. The round was led by Gradient, Google’s early-stage AI fund, with participation from General Catalyst, Liquid 2 Ventures, and Y Combinator.

The startup has also garnered support from leading AI figures, including François Chollet (Keras), Paul Graham (Y Combinator), Peter Wang (Anaconda), Michael Seibel (Twitch), Mike Knoop (Zapier), and Wes McKinney (Pandas). 

Haling from Hyderabad, 23-yr old Cyril Gorlla’s startup CTGT has also attracted funds from prominent investor Mark Cuban. 

“CTGT’s vision is to make AI more transparent and accessible without sacrificing on performance,” said Gorlla in an earlier exclusive interaction with AIM.

Challenging Deep Learning’s Inefficiencies

CTGT aims to address the growing inefficiencies in deep learning, a challenge that has persisted despite rapid advancements in AI models. Gorlla, who has long studied AI’s increasing demand for compute, believes that merely scaling models will not resolve their fundamental limitations.

Instead, CTGT has developed a new AI stack that transforms how models learn and train. The company claims its platform can customise, train, and deploy AI models up to 500 times faster than traditional methods, all while maintaining state-of-the-art accuracy. 

More importantly, this is achieved without requiring massive computational power, a significant departure from conventional deep learning approaches.

Enterprise Adoption 

CTGT’s AI deployment and quality platform is already in use by Fortune 10 enterprises, helping them gain more control over AI models in real-world applications. With fresh funding, the company plans to expand access to more enterprises looking to move AI from proof-of-concept to full-scale production.

Gorlla had told AIM that many existing AI methods remain computationally inefficient, citing an example where a leading foundation model provider’s state-of-the-art LLM interpretability requires more compute than the foundation model itself, making such methods inaccessible to most companies. 

By focusing on understanding the foundational mechanisms of learning, CTGT is developing AI models that are both efficient and interpretable. The company’s approach has been evaluated across multiple benchmarks. In a test involving 121 classification datasets, traditional neural networks required five hours for training, whereas CTGT’s method completed the process in just 40 minutes.

The company is now working on an upgraded training algorithm, which, according to Gorlla, will be 500 times faster than the current version.

“This is just the beginning of our journey in creating the next generation of truly intelligent AI: built from the ground up to be trustworthy and efficient, dynamically adapting to your needs,” he said in a Linkedin post.