TrueFoundry, a startup run by a group of former Meta engineers to help enterprises deploy AI systems at scale, has raised $19 million in fresh investment led by Intel Capital.
With GenAI’s emergence following the launch of ChatGPT in November 2022, enterprises of all sizes have looked at ways they could embrace AI. But GenAI relies on large language models to deliver intelligence (read efficiency), demanding significant compute power. Not all enterprises have the advantage of accessing hundreds or thousands of GPUs, making GPU optimization essential. Further, enterprises need to experiment with multiple models to find the best fit for their use case and fine-tune existing models to make AI systems relevant to their operations.
TrueFoundry addresses all this with its software available under a platform-as-a-service model, targeting full-stack data scientists. It provides features such as auto-scaling, proactive maintenance, centralized access controls, and real-time monitoring to simplify end-to-end AI deployments.
Founded in June 2021 by former Meta engineers Nikunj Bajaj and Abhishek Choudhary along with their IIT Kharagpur batchmate Anuraag Gutgutia, TrueFoundry initially offered a cross-cloud native software to speed up machine learning deployments. However, as GenAI became mainstream in 2023, the startup modified its system to support GenAI capabilities.
“Earlier, data scientists would only work with models and experiment with them, but a lot of the deployment was handed over to ML engineers. With our system, what we are enabling is that data scientists themselves can actually build this entire complex system and test it at scale before handing it over to the platform team for final deployments,” Gutgutia said in an exclusive interview.
The startup provides a software solution called an autopilot system, which uses AI to look at logs and metrics to help scale AI applications by adjusting GPU usage and fixing memory requirements.
Using the autopilot system, Gutgutia told TechCrunch, one of TrueFoundry’s customers with a small team of two people managed 10 million requests per second.
TrueFoundry currently has 30 paid customers worldwide alongside “a lot of” users accessing its open-source RAG framework for experimentation. Of these customers, GPU giant NVIDIA uses the software to build and deploy agents that optimize its GPU clusters. Med-tech company Resmed also uses TrueFoundry’s offering to develop an internal AI platform. Similarly, Siemens Healthineers, Automation Anywhere, Games24x7, and Whatfix are among the customers.
The startup has architectured its software using Kubernetes, which makes it multi-cloud ready and compatible with all major cloud platforms, including AWS, Google Cloud Platform, and Azure, among others. Additionally, enterprises can deploy the platform natively on-premise.
Enterprises using TrueFoundry have built and launched their internal AI platforms in two months and achieved their ROI in four months, unlike the industry average of 14 months, Bajaj stated. Gutgutia added that the startup delivered its customers with around 40-50% cost reduction on infrastructure spends, along with a 10× deployment speed improvement, no matter whether they are working on an agent, RAG, a complex deep learning model or a fine-tuned model.
Cloud vendors have started offering their own software to ease the building and deployment of AI models, such as Amazon’s Sagemaker and Google Cloud’s Vertex. Nonetheless, Gutgutia told TechCrunch that TrueFoundry’s solution is still pertinent as it helps “accelerate the usage of compute” for cloud providers, speed up the time-to-market for AI apps and reduce their overall costs.
The all-equity Series A round saw participation from Eniac Ventures, Peak XV Partners, and Jump Capital, as well as angel investors including Gokul Rajaram, Mohit Aron, and Cyan Bainster.
The startup will utilize the funding to expand the team from 45 members based in India to go-to-market positions in the U.S., including sales, customer success, and product marketing roles. It also plans to invest in partnerships, especially with cloud vendors and is in the process of getting listed in the key cloud marketplaces, beginning with AWS Marketplace.
Additionally, TrueFoundry aims to launch its AI agent that will help suggest the right resources and enable auto-scaling, as well as troubleshooting.
With the fresh funding, TrueFoundry has raised over $21 million in total, including the $2.3 million in its seed round in September 2022 led by Peak XV’s (then Sequoia India) Surge.
The startup quadrupled its customer base last year and deployed over 1,000 clusters across clients for ML workloads. Gutgutia also told TechCrunch that it is at “north of $1.5 million ARR” and is looking to double down this year.
