Artificial intelligence models have long struggled with long-term memory, the ability to retain and recall context across multiple sessions. Now, 19-year-old founder Dhravya Shah is tackling this challenge with Supermemory, a universal memory API designed to give AI applications lasting, contextual understanding.
Originally from Mumbai, India, Shah began his journey by building small consumer apps and bots. One of his early creations, a bot that formatted tweets into visually appealing screenshots, was sold to social media tool Hypefury. With proceeds from that sale, Shah moved to the U.S. to attend Arizona State University, where he challenged himself to create a new project every week for 40 weeks.
One of those projects became Supermemory, initially called Any Context, a tool that allowed users to chat with their Twitter bookmarks. Over time, Shah evolved the project into a full-fledged AI infrastructure product capable of extracting “memories” or insights from unstructured data, helping applications recall context over time.
The platform can process multiple data types, including files, chats, emails, PDFs, and app data, to build personalized knowledge graphs for each user. With its multimodal capabilities, Supermemory can even support video editors in fetching relevant assets or writers in recalling month-old entries. A Chrome extension and integrations with Google Drive, OneDrive, and Notion make it even more versatile.
“Our core strength is extracting insights from unstructured data and giving apps more context about users,” Shah said. “As we work across multimodal data, our solution suits AI apps of all kinds, from email clients to video editors.”
Supermemory has already raised $2.6 million in seed funding, led by Susa Ventures, Browder Capital, and SF1.vc, with angel investors including Cloudflare CTO Dane Knecht, Google AI chief Jeff Dean, DeepMind’s Logan Kilpatrick, and Sentry founder David Cramer.
Shah’s product has attracted an impressive roster of clients, such as AI video editor Montra, AI search platform Scira, and a16z-backed Cluely. It’s also collaborating with a robotics firm to help machines retain visual memories.
Related: AI’s Reinforcement Gap: Why Coding Tools Are Evolving Faster Than Writing Assistants
Despite rising competition from other startups like Letta, Mem0, and Memories.ai, Shah believes Supermemory will stand out due to its low latency and high performance.
“More and more AI companies will need a memory layer,” said Joshua Browder, founder of DoNotPay and one of Supermemory’s investors. “Supermemory provides high performance while surfacing relevant contexts quickly.”
As AI developers continue to push the boundaries of context retention, Shah’s Supermemory could be the key to building truly “long-term thinking” machines.
