
Sign up to save your podcasts
Or


When $300B gets wiped off SaaS market caps, it’s usually not just sentiment, it’s the market sensing a platform shift. AI agents are replacing per-seat SaaS software, vector databases are becoming the memory layer for large language models (LLMs), and knowledge infrastructure is emerging as the real foundation of enterprise AI. In this episode, Ash Ashutosh, CEO of Pinecone, explains why LLMs need a new database category, why traditional databases break in the age of semantic search, and how vector databases are powering the next generation of enterprise AI.
Inside the episode:
* Why production AI requires purpose-built memory infrastructure
* How Pinecone grew to 9,000+ paying customers and 800,000+ developers
* The shift from experimentation to enterprise-scale deployment
* How token costs and context management are becoming core design constraints
* Why the $300B SaaS wipeout signals a structural shift
* The death of per-seat pricing in an agent-driven world
* Why LLMs are powerful and yet fundamentally broken without memory
* The four defects of LLMs
* Why traditional SQL and document databases fail under semantic search
* Vector databases explained simply: storing meaning, not rows
* Why multi-agent systems make 96% accuracy dangerously insufficient
* Where the next $100B AI companies might actually be built
Watch full episode on YouTube:
About Ash Ashutosh:
Ash Ashutosh is a multi-time founder who has repeatedly built and defined entirely new data categories. He founded AppIQ in 2001, pioneering storage resource management before its acquisition by HP. He later built Actifio, creating the copy data management category, which was acquired by Google. Today, as CEO of Pinecone, Ash is leading one of the most important infrastructure shifts in AI, building the vector database layer powering enterprise large language models and agentic systems. Few operators have shaped as many foundational layers of the data stack and fewer still are doing it again at the dawn of the AI era.Converge VC website: https://converge.vc/Connect with Ash Ashutosh: https://www.linkedin.com/in/ashashutosh/Connect with Nilanjana Bhowmik: https://www.linkedin.com/in/nilanjanabhowmik/
By In-depth conversations with top founders and VCs on building, scaling, and raising capital across industries.When $300B gets wiped off SaaS market caps, it’s usually not just sentiment, it’s the market sensing a platform shift. AI agents are replacing per-seat SaaS software, vector databases are becoming the memory layer for large language models (LLMs), and knowledge infrastructure is emerging as the real foundation of enterprise AI. In this episode, Ash Ashutosh, CEO of Pinecone, explains why LLMs need a new database category, why traditional databases break in the age of semantic search, and how vector databases are powering the next generation of enterprise AI.
Inside the episode:
* Why production AI requires purpose-built memory infrastructure
* How Pinecone grew to 9,000+ paying customers and 800,000+ developers
* The shift from experimentation to enterprise-scale deployment
* How token costs and context management are becoming core design constraints
* Why the $300B SaaS wipeout signals a structural shift
* The death of per-seat pricing in an agent-driven world
* Why LLMs are powerful and yet fundamentally broken without memory
* The four defects of LLMs
* Why traditional SQL and document databases fail under semantic search
* Vector databases explained simply: storing meaning, not rows
* Why multi-agent systems make 96% accuracy dangerously insufficient
* Where the next $100B AI companies might actually be built
Watch full episode on YouTube:
About Ash Ashutosh:
Ash Ashutosh is a multi-time founder who has repeatedly built and defined entirely new data categories. He founded AppIQ in 2001, pioneering storage resource management before its acquisition by HP. He later built Actifio, creating the copy data management category, which was acquired by Google. Today, as CEO of Pinecone, Ash is leading one of the most important infrastructure shifts in AI, building the vector database layer powering enterprise large language models and agentic systems. Few operators have shaped as many foundational layers of the data stack and fewer still are doing it again at the dawn of the AI era.Converge VC website: https://converge.vc/Connect with Ash Ashutosh: https://www.linkedin.com/in/ashashutosh/Connect with Nilanjana Bhowmik: https://www.linkedin.com/in/nilanjanabhowmik/