Building the Trust Layer for an Agent-Driven World
If you’ve spent time here at Venture Dock, you’ve likely met Arjun Subedi, a founder and community-builder whose work sits at the intersection of deep-tech research and real-world company building. Originally from Nepal, Arjun spent years in Chicago before moving to Silicon Valley in search of a place where industry-defining startups actually take shape. “To have asymmetric growth, you need to be around heterogeneous people,” he told us. “People who are amazing in different fields, that’s how you connect dots.”
That instinct is what led him into bouncing between post-doc paper sessions and technical meetups. He quickly noticed something missing: “A lot of PhDs had never even been asked, ‘Have you thought about starting a company around this?’” What began as curiosity turned into a network, later the foundation of The Deep-Tech Community, which Arjun helped shape long before his newest venture.
From Community to Company
Arjun’s latest startup, Astha.ai, was born from patterns he saw repeatedly: a wave of AI agents emerging… with shockingly little conversation about risk or safety. “Everybody was optimistic about agents, but nobody was talking about security,” he said. “Security becomes the ultimate bottleneck.”
The team’s first product, which is an identity layer for agents, was too early. “It was like building a prison without prisoners,” Arjun joked. That pivot led them to something bigger: SAFE-MCP, a governance and security framework now supported by both the OpenID Foundation and Linux Foundation. “Trust isn’t technology, it’s a feeling,” Arjun said. “Companies need guidance, not just tools.”
SAFE-MCP now helps enterprises shrink six-month security reviews into weeks and gives compliance-heavy sectors like fintech, healthcare, and biotech a structured way to adopt agents without introducing catastrophic risk.
A Future Defined by Agents
Arjun sees a near future where every app is an agent and much of daily life runs in the background. “In 10 years, everything will be agentified,” he explains. That shift brings enormous risk: agents persuading other agents, financial actions triggered without human awareness, and attacks happening at millisecond speed. “Traditional security companies aren’t thinking about this. We need something like laws, a constitution, for agents.”
Astha.ai aims to become that safety and governance layer.
Why Community Still Matters
For Arjun, community has always been the unfair advantage. “Startup life is an emotional roller coaster, way more downs than ups,” he said. “Having a community really helps.” Venture Dock, for him, has become a home base. “I’ve done 13–14 hour sprints here, sometimes until 1 AM,” he laughed. “It’s a place where you can get work done and be around people who understand the journey.”
His advice: show up consistently, contribute, and find your own way to create value. “Community isn’t transactional. It’s a long-term asset.”
About Astha.ai & The Deep-Tech Community
Astha.ai is building the trust, safety, and governance layer for the agentic era. Through SAFE-MCP, Astha.ai provides the industry’s first standardized framework for agent attacks, mitigations, and deployment guidance, helping organizations adopt AI agents securely and faster than traditional review cycles allow.
Alongside the company, Arjun continues to anchor The Deep-Tech Community, a network of researchers, builders, and VCs bridging the gap between academia and industry. What began as Arjun attending paper discussions and asking founders-in-waiting the simple question “Why not build a company around this?” has grown into a structured, collaborative ecosystem supporting deep-tech founders across the Bay Area.
Together, Astha.ai and The Deep-Tech Community reflect Arjun’s mission: to push the frontier of safe, intelligent systems while cultivating the environment that makes true deep-tech innovation possible.
