On the latest AZTF News stream, we covered one of the most compelling stories in Arizona tech right now: a young founder building foundational AI infrastructure from his dorm room at Arizona State University — and now shipping out of San Francisco
That founder is 19-year-old Dhravya Shah, and his startup is called Supermemory. It’s an API platform that gives AI systems a long-term memory layer, allowing chatbots and assistants to retain and recall user context across conversations, apps, and even platforms. If ChatGPT is your brain, Supermemory is your hippocampus.
“Context is everything,” Shah’s site reads. “Without it, even the smartest AI is just an expensive chatbot.”
Arizona Roots, Global Ambitions
Supermemory was born right here in Arizona. Shah started building what would become Supermemory while studying at ASU, originally launching a consumer-facing notes and bookmarks app called AnyContext. Friends told him it was a waste of time — but he kept building.
That decision paid off. As he shared on the stream, the real pivot happened when he started posting architecture diagrams online. Developers and companies began reaching out asking if they could just license the backend. The shift to a developer-first API was born. Shah rebranded the product Supermemory and turned it into infrastructure.
Shah also co-founded DevLabs at ASU, a student-led engineering group, where other talented builders have emerged. While not a cofounder of Supermemory, his collaborator Dhanush Vardhan (of DevLabs) remains part of the early circle of Arizona builders pushing the space forward.
Though Shah is currently based in San Francisco, he’s staying close to his Arizona roots — and plans to return after the summer. On the stream, AZTF host Mat Sherman called him “one of the most high-signal builders I’ve ever met. He’s cracked, but with zero ego.”
What Supermemory Does
Supermemory is a developer API that helps any AI product — from chatbots to agents to customer support tools — store, search, and recall user context. Instead of cramming the entire conversation history into every prompt (which is slow, expensive, and error-prone), Supermemory acts as a sidecar memory layer.
Developers send content — text, links, PDFs, notes, videos, or CRM data — to Supermemory’s API. Then when the AI needs to respond to a prompt, it queries Supermemory first for relevant context. Think: “What’s my name again?” or “What did I say about product strategy last month?”
According to their developer docs, Supermemory supports:
- Cross-platform memory syncing
- Integrations with Notion, Google Drive, and CRMs
- Low-latency search at scale
- On-prem, cloud, and device-based storage options
- Multi-model compatibility (ChatGPT, Claude, Gemini, etc.)
- SDKs for Python and JavaScript
In short, it’s trying to become the Stripe of AI memory.
Early Buzz and Community Support
In April 2025, Shah launched Supermemory MCP (Memory Context Protocol), a cross-LLM memory plugin that lets users carry context across models like ChatGPT, Claude, and Gemini. Weeks later, during Y Combinator’s AI Startup School, OpenAI CEO Sam Altman announced MCP live on stage as a key building block for future AI agents — but by then, Shah had already shipped it. In fact, as Altman demoed the idea, Shah was sitting in the audience building the ability for their users to interface with OpenAI’s MCP.
The project has already seen early traction:
- Over 60,000 users from Shah’s original app
- Thousands of developers using the API
- Product of the Day on Product Hunt
- Integration wins, like with AI notetaking app Flow
On our stream, co-host Phillip Pipkins described it as a “sidechain for memory — so the LLM doesn’t have to waste tokens reprocessing everything. It just pings Supermemory and keeps going.”
The tech community seems to agree. From hackathon wins at HackPrinceton to building open-source memory layers, Shah has stayed close to the builder ethos that defines the best founders.
What’s Next
So far, Supermemory hasn’t raised a big round. It’s funded by angels, including Sherman, who revealed live that it was the first company he ever wrote a check into. And that early bet is already aging well.
With more startups, agents, and AI tools realizing that memory is the key to personalization, Shah is betting big on making Supermemory the infrastructure layer behind it all. “Everyone is adding their own memory layer,” the website reads. “Why not carry it with you?”
If Arizona is going to be known for not just launching startups, but building real technical infrastructure, Supermemory might be the blueprint — quietly powerful, deeply useful, and rooted in the desert.