Denise Persson’s Post

Snowflake just unleashed Arctic, a cutting-edge Enterprise LLM that's revolutionizing data integration! Experience it firsthand by visiting https://lnkd.in/gvBSD7VE and witness the magic in action. Ready to dive in? Test its incredible capabilities with queries like "How to survive my wife and kids," "How to be a faster cyclist," or "Summarize the current events in [Insert region]." Join the data revolution with #Snowflake and #Arctic!

Its not every day that a new AI model gets as much business attention as Snowflake Arctic generated. ❄️ For Blend clients and AI enthusiasts wondering about the impact Arctic will have on the LLM scene, Rob Fuller, SVP of Technology Solutions at Blend, and Luke Turanski, Lead Data Engineer at Blend, put together an introductory guide. For those of you in a rush, here are the quick facts: 🏢 Arctic’s open-source materials is an innovative step-forward for Enterprise LLMs 📈 Outstanding scores in SQL generation, coding, and more 🏗️ Arctic will be built into Snowflake Cortex’s new features Snowflake’s announcement of Arctic comes alongside adjacent features like Streamlit in Snowflake, Snowflake ML, and Snowflake Container Services– more than enough to get any AI shop excited (We sure are). Learn more about how you can leverage Arctic in your AI-problem solving stack: https://lnkd.in/gsjx6Jfr #AIModels #LargeLanguageModels #LLMs #SnowflakeArctic #EnterpriseLLM

📢 Snowflake launches its own open-source LLM. 💡 Snowflake has introduced the Arctic, an innovative language model crafted by Snowflake AI Research. The Arctic aims to transform our communication and collaboration methods, promising unmatched efficiency, intelligence, and transparency in interactions. 📈 1. Enterprise Intelligence: Arctic excels in complex enterprise tasks like SQL and code generation, setting new benchmarks in instruction following and more. 👨💻 2. Breakthrough Efficiency: Arctic's dense-MoE hybrid architecture yields top results economically. 👾 3. Truly Open: Arctic: Apache 2.0 license, open weights and code, plus data recipes and research insights. Check it now: Snowflake Arctic https://lnkd.in/dyD7p2kH #Snowflake #OpenAi #GenAI #AI #Arctic #AI_Research #LLM

Breaking down development silos is the next frontier for Snowflake. Using #Snowpark, #Cortex and now #Arctic you can bring the work safely to the data.

While going through DB indexing, I came across OpenAI's acquisition of Rockset. Curious! I dug deeper and was amazed to learn about their real-time indexing approach. They combine different indexing techniques as soon as data is ingested, making it immediately ready for optimized queries. Key Takeaways: 1. Real-time indexing: Query ready data upon ingestion 2. Converged Index: Combines multiple index types 3. Cloud-native architecture 4. Compute-Compute Separation: Isolates ingestion/indexing and querying 5. Hybrid search: Optimized for vector, text, and metadata Original whitepaper: https://lnkd.in/eshnHCuP

Today is a momentous day for Snowflake! We have launched Snowflake Arctic, an open-source large language model that is designed to be the most open and enterprise-grade LLM available in the market. This state-of-the-art language model is uniquely designed to be the most open, enterprise-grade LLM on the market, and we're excited to see the impact it will have on our customers. Check out the link to learn more about Snowflake Arctic: https://lnkd.in/g-KCEKKt #Snowflake #LLM #OpenSource #EnterpriseGrade

Exciting development! OpenAI's acquisition of Rockset truly showcases innovation in real-time indexing and data query optimization. Their approach to converged indexing, cloud-native architecture, and hybrid search capabilities is setting a new standard for efficiency and scalability in data management. Looking forward to seeing the impact of these advancements in enhancing real-time analytics and applications across industries! #RealTimeAnalytics #IndexingInnovation

The OpenSauced platform has TONS of lightning fast time-series and histogram visualizations: any one given page has a huge number of charts, each with their own array of data. How do we handle serving all this data and what makes it all so fast? I recently wrote a piece for Timescale's blog on how we use their database service for ingesting GitHub events, structuring terabytes of data, serving very fast endpoints for our visualizations, and empowering our AI enabled features. TLDR: Timescale is fast, very appealing for our small team of engineers that already know Postgres, and provides some awesome features right out of the box that enabled us to build StarSearch, our first huge scale AI product. Very excited for where Timescale has brought us and where it will take us! #database #ai #timeseries #opensource

Snowflake Cortex went GA today and it's a game-changer... 😱 Cortex lets you: Bring leading models from AI at Meta and Mistral AI to your data in Snowflake without managing the underlying compute. Use serverless SQL or Python functions to analyze text data, or generate predictions and insights on your structured data with ML-based functions. Quickly build context-enriched applications using Retrieval Augmented Generation (RAG), as well as SQL and Python serverless functions for vector embedding and similarity search. Turn text to SQL with a breakthrough AI-powered assistant. Extract values from unstructured data with Snowflake Arctic. Worth restating: Since these LLMs are fully hosted and managed by Snowflake, using them requires no setup. Your data stays within Snowflake, giving you the performance, scalability, and governance you expect. I'm so excited to see what our customers build with this. 👀 Sridhar Ramaswamy says it's the year of GA and he is not messing around. #technology #innovation #data #sales

I am beyond excited to announce the partnership launch between Weaviate and Snowflake Snowpark Container Services!  ❄ Snowflake customers can now leverage open-source Weaviate + open-source LLM’s of choice to securely deploy enterprise-grade AI applications like Generative Feedback Loops, Semantic Search, Large Scale RAG, and more.  Check out this blog post from my colleague (and fellow ex-Snowflake) Jon Tuite where he demonstrates how, in 3 hours time and less than $60 in infrastructure cost (w/ NVIDIA GPU), you can build a Generative Feedback Loop for product review data.  The demo includes 6 steps: 1. Loaded Product Review data into Snowflake 2. Deployed Weaviate, Ollama, and a vectorizer within SPCS 3. Created product metadata using the latest Mistral model 4. Loaded a new product table into Weaviate 5. Performed semantic and hybrid searches using Weaviate 6. Kept all data within Snowflake https://lnkd.in/eMUCBHwF This is going to provide tremendous value for Snowflake customers.  Please reach out to me for more info. To the moon! 🚀 🚀 🚀 #weaviate #spcs #snowpark #generativeAI #vectordatabase #LLM https://lnkd.in/e9F_PsuS