Browserbase reposted this
What areas in the data layer have the most immediate opportunity to be re-invented due to AI? 🤖 We’re seeing innovation across: •unstructured data extraction and pipelining •RAG •data curation •data storage •AI memory Check out our Felicis deep dive at the link in the comments Eric Flaningam #data #AI #infrastructure
This entire map can be compressed in a single product: StackAI
There are a lot of "modern data stack" tools in here. They've solved the data acquisition problem for the SQL-focused data / analytics engineer. For AI, we need to think about Python developers, and how we solve the data acquisition problem for that part of the ecosystem. dltHub is one example.
Weird to not see Airbyte in ingestion given we’re more advanced (in terms of unstructured sources and vector databases) for AI use cases…
Sam Altman did a podcast with All-In a few weeks ago where he called out how important data will be at time of inference to do something useful. Bullish on that approach, so we’re betting on it. Currently, we’re building agents & chains for secondary research workflows that synthesize data from various third party sources at runtime and deliver professional, consulting-like outputs in minutes.
Astasia, your exploration of AI-driven innovations in the data layer is fascinating. Reflecting on your insights, it's clear that unstructured data extraction and AI memory are set to redefine how we approach data management. How do you foresee these shifts influencing enterprise strategies in the near future?
Astasia Myers I’d like to understand more about what is meant by lineage. What seems to be missing in most AI training data workflows us provenance—proof of authenticity, attribution, legal right to use (owned or licensed)—before the data gets ingested for AI model training. #proveableprovenance #ethicalai #aidataengineering
Great map! Thanks for sharing. I would also add Raito in the data security space. Raito helps data platform teams monitor, manage, and automate data access and usage across the data stack.
Looks more like brand catalog across the stack . 😎
General Partner at Felicis
1mohttps://www.felicis.com/insight/ai-data-infrastructure