Nineteen-year-old founder Dhravya Shah has secured significant backing from tech heavyweights, including Google AI Chief Jeff Dean, for his startup, Supermemory. The company is building a crucial memory solution for AI applications, an area where current large language models (LLMs) often struggle to maintain context across multiple sessions.
From IIT Prep to Silicon Valley Seed Round
Originally from Mumbai, India, Shah started his journey by building and selling consumer-facing bots. He successfully sold one bot that formatted tweets to the social media tool Hypefury, earning enough to change his path from preparing for the rigorous IIT entrance exam to moving to the U.S. and attending Arizona State University.
After relocating, Shah challenged himself to launch a new project every week for 40 weeks. During this period, he created the initial concept for Supermemory (originally called Any Context), which first allowed users to chat with their Twitter bookmarks.
After securing an internship at Cloudflare in 2024, where he worked on AI and infrastructure, advisors like Cloudflare CTO Dane Knecht encouraged him to turn Supermemory into a full-fledged product. Shah decided to pursue the startup full-time this year.
A Universal Memory API for AI
Supermemory is now described as a universal memory API for AI apps. Its core function is to extract “memories” or insights from unstructured data to build a knowledge graph, helping applications better understand and personalize context for users.
This solution allows for deep contextual querying across vast amounts of data. For example:
- A writing or journaling app could query entries that are months old.
- An email client could search through a massive archive of messages.
- The solution’s multimodal input means a video editor could use it to fetch relevant assets from a library based on a simple prompt.
Supermemory can ingest virtually any type of data, including files, chats, emails, PDFs, and app data streams. It also offers a chatbot and a notetaker feature for users to easily add new memories, files, or links, and can connect directly to apps like Google Drive and Notion.
Shah emphasizes the company’s unique value: “Our core strength is to extract insights from any kind of unstructured data and give the apps more context about users.”
Securing $2.6 Million from Industry Leaders
Supermemory has closed a $2.6 million seed funding round led by Susa Ventures, Browder Capital, and SF1.vc.
The round included an impressive roster of individual investors, underscoring the demand for a dedicated memory layer:
- Jeff Dean (Google AI Chief)
- Dane Knecht (Cloudflare CTO)
- Logan Kilpatrick (DeepMind Product Manager)
- Executives from OpenAI, Meta, and Google
Joshua Browder, founder of DoNotPay who invested via Browder Capital, was drawn to Shah’s speed: “What struck me was how quickly he moves and builds things, and that prompted me to invest in him.”
Supermemory is already working with multiple customers, including the a16z-backed desktop assistant Cluely, and is even partnering with a robotics company to help a robot retain its visual memories.
While the memory space is competitive, with players like Letta and Mem0 also building memory layers, Shah is confident Supermemory will stand out due to its lower latency and high performance in surfacing relevant context quickly.
As Browder noted, “More and more AI companies will need a memory layer. Supermemory’s solution provides high performance while allowing you to surface relevant context quickly.”
