Filip Vítek

AtScale + Snowflake + Delphi Labs - Unlock the Potential of Self-Service Analytics – https://bit.ly/3UOjIBu The Future of Self-Service Analytics with Snowflake, Semantic Layers & LLMs - How to Deliver Self-Service Analytics on Snowflake Using a Semantic Layer and LLMs Join AtScale’s webinar on May 22nd to unlock next-level data analytics with Snowflake, AtScale, and Delphi Labs. Learn from industry experts how to integrate Snowflake's powerful platform with AtScale's Semantic Layer and leverage LLM’s for transformative insights. Register today! Wednesday, May 22, 2024 2:00 PM ET (11:00 AM PT) | 60 mins Featured Speakers • Dave Mariani, Founder & CTO, AtScale • Alex Ross, Senior Solutions Engineer, Snowflake • David Jayatillake, CEO, Delphi Labs • Michael Irvine, Co-Founder, Delphi Labs You’re invited! - https://bit.ly/3UOjIBu Learn how to integrate Snowflake's data platform with AtScale’s Semantic Layer and Large Language Models (LLMs) to enhance self-service analytics capabilities. You'll walk away with actionable insights to harness the full potential of this innovative approach. Don't miss this opportunity to enhance your data strategy and stay ahead in today's dynamic data landscape. In this webinar, we’ll cover: • The Foundation of Data Management: Learn from Alex Ross, Senior Solutions Engineer at Snowflake, about the fundamentals of an efficient data platform. • Introduction to Semantic Layers: Discover how a semantic layer can simplify data accessibility and enhance business insights with Dave Mariani, CTO at AtScale. • Power of LLMs in Analytics: Delphi Labs will present a live demonstration of how LLMs integrated with a semantic layer can revolutionize data interaction. Who should watch This webinar is designed for those eager to streamline their analytics processes and make data-driven decisions faster and more efficiently.

2

You know a Semantic Layer would be hugely valuable, but how do you actually build such a thing? Pro tip: Crafting a Semantic Layer is about building iterative velocity alongside accuracy, so that when your stakeholders ask about Revenue MoM grouped by Attribution Channel, you can answer instead of adding a ticket to the backlog. Start with these four steps: 1. Identify a Data Product that is impactful: Find something that is in heavy use and high value, but fairly narrow scope. Don’t start with a broad executive dashboard that shows metrics from across the company because you’re looking to optimize for migrating the smallest amount of modeling for the highest amount of impact that you can. For example, a good starting place would be a dashboard focused on Customer Acquisition Cost (CAC) that relies on a narrow set of metrics and underlying tables that are nonetheless critical for your company. 2. Catalog the models and their columns that service the Data Product, both in dbt and the BI tool, including rollups, metrics tables, and marts that support those. Pay special attention to aggregations as these will constitute metrics. 3. Melt the frozen rollups in your dbt project, as well as variations modeled in your BI tool, into Semantic Layer code. 4. Create a parallel version of your data product that points to Semantic Layer artifacts, audit, and then publish. Creating in parallel takes the pressure off, allowing you to fix any issues and publish gracefully. You’ll keep the existing Data Product as-is while swapping the clone to be supplied with data from the Semantic Layer. Dig deeper into the step-by-step process of how to ship a Semantic Layer in pieces at our link in the comments.

371

25 Kommentare

Snowflake Cortex is now available to Snowflake customers, and you may have thought - yawn - another big software company has launched Generative AI capabilities. But this is actually a game-changer for marketers. With Snowflake's Cortex now in place, brands can use Google AI, Mistral AI, Meta AI, Snowflake's own Artic, and other leading generative AI products in a secure, efficient, and user-friendly way... #CDP #customerdataplatform #snowflake

17

1 Kommentar

Wake up! LLMs have jolted companies from their data slumber. But mountains of disjointed info, scattered silos, ignored guidelines, and skyrocketing costs are getting in the way. Sanjay Deshmukh tells us how Snowflake is stepping up as the new data sheriff and bring order to this new chaotic data frontier. #data #snowflake #LLM #generativeai #AI #datamanagement

14

Why Data Quality is Non-Negotiable for Data Monetization? In order to succeed in the world of data-driven decisions, data quality reigns supreme. Without accurate, consistent, and complete data, your insights falter, your decisions misfire, and your monetization efforts fall flat. Here’s why data quality matters: ✅ Better Decisions: Accurate data powers smarter insights and choices. ✅ Trust & Credibility: Reliable data builds confidence with customers, partners, and investors. ✅ Revenue Potential: High-quality data commands higher value in the market. ✅ Cost Savings: Avoid costly mistakes and inefficiencies with accurate, trustworthy data. Want to elevate your data game? Start with these steps: ✅ Assess Your Data: Start with a data quality assessment to understand its structure and consistency. ✅ Data Hygiene: Correct inaccuracies and fill in missing gaps. ✅ Standardize Formats: Keep your data structured and uniform. ✅ Audit & Validate Regularly: Audit and use rules to ensure ongoing accuracy. ✅ Data Governance: Ensure you policies in place to maintain quality over time and ensure privacy and compliance. Investing in data quality is a strategic imperative for unlocking growth and maximizing the value of your data assets. How does your organization ensure data quality? Let’s discuss! 👇 💬 Drop your thoughts in the comments, and if you find this content useful, follow me for more insights like this. #DataQuality #DataMonetization #BigData #DataDrivenGrowth

11

7 Kommentare

Databricks vs. Snowflake: What their rivalry reveals about AI's future - by Ashu Garg of Foundation Capital https://lnkd.in/eBJ-TAnu Other nuggets on the market potential for enterprise AI: 1. 80-90% of enterprise data is unstructured data 2. 89% of a survey of 1400+ execs from Fortune 500 rank genAI in their top 3 priorities 3. AI could increase global GDP by $13-15.7T by 2030 I agree with Ashu that there's room for both "scale maximalists" and "small but mighty" in the model wars. In our experience deploying generative AI at Fortune 500 at Turing, we see that businesses need powerful applications that bring together a) Models b) proprietary data c) proprietary human knowledge (still sitting in the minds of human experts) to make generative AI sing and automate complex workflows. I also believe most systems will have some human in the loop and these systems will amplify human productivity. We are still in the early days of an exciting journey ahead! :) 🤖

152

4 Kommentare

If you use the right predictive model, you can have an outsized influence on the "movable middle." Here's how our approach is moving the needle for our clients: 1. Model Selection: We're flexible. Whether it's a regression model or a sophisticated machine learning algorithm, we choose what works best for the specific challenge at hand. 2.Customer Ranking: Our models rank customers from most to least likely to exhibit a desired behavior. 3. Decile Division: We split these rankings into 10 equal groups, or deciles. From there, we can lump customers into one of three categories: -Top Deciles: They'll likely engage without extra incentives. -The Middle: This is often where the greatest opportunity lies. -Bottom Deciles: Unlikely to engage regardless of efforts. We focus on that potentially movable middle group. By testing different offers, messages, and channel mixes, we can influence behavior where it matters most. By understanding where customers fall in these deciles, we can optimize marketing efforts and allocate resources for maximum impact. #DataDrivenMarketing #CustomerInsights

9

1 Kommentar

I'm a huge advocate of the power of data warehouses and how they can completely transform businesses. Here's why: 🔑 A single source of truth A DWH pulls in marketing, sales, and customer service data, becoming your source of truth. ⚙️ Data Activation With a DWH, you can actually use all that data across multiple platforms according to your business needs—whether it’s better targeting on Google Ads, personalizing email marketing, or using Tableau for advanced insights. 📈 Expansion Opportunities Now that your data is in one place, it’s easier to test new initiatives like mobile apps or automation software. You can also explore cohort analysis, calculate lifetime value, and build customer retention models, getting the insights you need to make strategic decisions. 🛡️ Risk Mitigation Vendor lock-in – being tied to one provider and having limited flexibility is a big risk many businesses face. A DWH gives you more freedom to scale and switch vendors without losing access to your data. 🔍 Business Intelligence You can layer tools like LookerStudio on top of your DWH to uncover insights like product sale-through rates, high abandoned-cart issues, and customer behavior trends. 🔒 Data Security Instead of sending CSVs or customer data over email (which isn’t ideal), you can securely share data with third parties using endpoints, while also anonymizing any information you don’t want to disclose. 🎯 Expand to Third-Party Integrations & Predictive Analytics With a DWH, you can start layering in third-party data like age, income, and device usage to build more detailed customer profiles. From there, you can tap into predictive analytics for insights like: Predictive LTV (lifetime value) Predictive churn Predictive fraud This unlocks a deeper level of personalization and allows you to test personalized product recommendations and website promotions. 💰 Inventory and Financial Forecasting You can even integrate your financial and inventory data for a full picture of your business. With a DWH, you can predict sales and inventory trends, helping your team make smarter decisions.

1

How startups in Data/AI/ML build their GTM teams? Insights from 500,000+ employees reveal a consistent pattern. We've analyzed data from over 500,000+ current and former employees across more than 1,000 AI/ML companies. Here are the key findings: 1) Initial focus 60% of the first GTM hires were in sales and support, followed by 37.5% in marketing, and just 2.5% in engineering. (*Note: engineering in this case would refer to solutions/sales engineers) This makes sense. Sales is the obvious first stop when one thinks about selling a product. BUT, an interesting trend we've seen in recent years is a commitment to standing up the marketing flywheel EARLY. More and more companies are pledging resources to brand, content, and product marketing very early on (hence the 37.5% slice here). 2) Expanding the team: Subsequent hires increasingly favor sales and support, with 80% for the second hire and 90% for the third, emphasizing the critical role of building strong sales teams. ***Here's where I think this data can be slightly deceiving: We capture data across companies that range from seed stage all the way to industry Goliaths like Snowflake. Yes, all these companies were startups at one point. BUT, building a GTM function in 2024 is not like it was in 2010. The motion back then, an thus the personnel needed, are different. We are forgetting about the very first GTM team member: The Founder! In the early days, the founder is the seller. Some other things to consider - 🤌Marketing typically takes a longer time to marinate. Unless you are crushing a paid media lead capture approach (which is less likely to be effective in today's market), founders should expect marketing to be a longer term investment. ☎️Sales might lead you to quicker results, but headcount costs can rack up quickly. 🕵️The very first GTM hires (whether sales or marketing) will need to to focus on more than just immediate revenue. They will: a) help you refine the product, b) help you define your ICP, c) help you design a GTM motion that resonates with your prospects. Thus, these are special people. They are special hires. Check out these two posts I made that breakdown the activities of early GTM hires at Snowflake and Hubspot: Hubspot: https://lnkd.in/gTCzxfZr Snowflake: https://lnkd.in/gyMzVQez Thanks to Jason Saltzman from Live Data Technologies for the crucial employment data (visit livedatatechnologies.com for more). Interested in a deeper dive? DM 'DATABASE' for a customizable data export. With this database's support, we've assisted numerous successful early-stage startups—backed by giants like Insight Partners, Sequoia Capital, Andreessen Horowitz, and more—in securing top-tier talent.

49

10 Kommentare

We overhauled the dashboard experience in Causal and I am loving the new look 🤩 The new layout is way more flexible and configurable to let users decide the best way they want to showcase their data. In the new dashboard layout: 🎯 Users can add new visuals directly in the position you want them (no more dragging from the bottom of the dashboard!) 💬 Text is more native, allowing you to break up your dashboard into digestible sections, and tell the story you want alongside your data 3️⃣ You can now have up to three visuals on one row (vs. two before), which will be especially useful for charts, KPIs, or side-by-side text blocks underneath visuals. 〰️ With two visuals on a row, you can now have a one-third: two-thirds layout (instead of just half & half), giving you more room for visuals that need space to breathe 🐭 Changing the position of visuals with drag and drop should feel a lot smoother + more responsive.

71

10 Kommentare

AtScale + Snowflake + Delphi Labs - Unlock the Potential of Self-Service Analytics – https://bit.ly/3UOjIBu The Future of Self-Service Analytics with Snowflake, Semantic Layers & LLMs - How to Deliver Self-Service Analytics on Snowflake Using a Semantic Layer and LLMs Join AtScale’s webinar on May 22nd to unlock next-level data analytics with Snowflake, AtScale, and Delphi Labs. Learn from industry experts how to integrate Snowflake's powerful platform with AtScale's Semantic Layer and leverage LLM’s for transformative insights. Register today! Wednesday, May 22, 2024 2:00 PM ET (11:00 AM PT) | 60 mins Featured Speakers • Dave Mariani, Founder & CTO, AtScale • Alex Ross, Senior Solutions Engineer, Snowflake • David Jayatillake, CEO, Delphi Labs • Michael Irvine, Co-Founder, Delphi Labs You’re invited! - https://bit.ly/3UOjIBu Learn how to integrate Snowflake's data platform with AtScale’s Semantic Layer and Large Language Models (LLMs) to enhance self-service analytics capabilities. You'll walk away with actionable insights to harness the full potential of this innovative approach. Don't miss this opportunity to enhance your data strategy and stay ahead in today's dynamic data landscape. In this webinar, we’ll cover: • The Foundation of Data Management: Learn from Alex Ross, Senior Solutions Engineer at Snowflake, about the fundamentals of an efficient data platform. • Introduction to Semantic Layers: Discover how a semantic layer can simplify data accessibility and enhance business insights with Dave Mariani, CTO at AtScale. • Power of LLMs in Analytics: Delphi Labs will present a live demonstration of how LLMs integrated with a semantic layer can revolutionize data interaction. Who should watch This webinar is designed for those eager to streamline their analytics processes and make data-driven decisions faster and more efficiently.

4

What parts of the data model should be built in dbt? in the semantic layer? in the last-mile layer? The (many) posts about this usually make the same mistake. They only include the first two layers and skip the last-mile. That's like writing a recipe for a loaf of bread and skipping the baking instructions. Data people like to pretend like the last-mile doesn't exist. Because the last-mile often happens in excel or sheets, and our egos can't accept that. But the reality is our structured data pipelines have a trade-off: reduced flexibility. The end-users need their data in a format that can't be expressed by a structured pipeline. So (officially or unofficially) the last-mile happens. Let's break these down, honestly, without pretending the last-mile layer doesn't exist: 1) The transformation layer does the heavy lifting. It's a series of SQL steps for cleaning, combining, and normalizing data. The goal is a set of tables at the end of the pipe exposed for analytics. You should do as much as possible here because it's pre-computed (and the more you pre-compute, the faster your data tools will be). This is all done in SQL with a tool like dbt, Coalesce, or SQLMesh. 2) The semantic layer happens at query-time. This is where you'll dynamically slice, aggregate, and filter data - anything that would create redundant tables in the transform layer. For example, the semantic layer aggregates an orders table into daily/weekly/monthly sales (if you did this in the transformation layer, you'd need to point your users at 3 redundant tables). The semantic layer is built inside the BI tool, or using a standalone semantic layer such as Cube or Metricflow. 3) The last-mile layer is the one everyone neglects. The last-mile includes all steps after the query is run. Taking the big, tall table prepared by the semantic layer, and turning it into the right shape for your project. That includes things like pivoting, formatting, or certain additional calculations (percentage growth is the classic example). Sometimes you'll combine another query. This is all done in excel (the horror!!), or as table calcs/merged results in the BI tool. Each layer has its purpose. The transform layer prepares a clean, consistent dataset. The semantic layer handles aggregation and filtering. And the last-mile layer gets data into the desired format. As a general rule, you should push everything as far left as you can, but not farther. Why am I writing this now? Because we released our great new merged results feature in Zenlytic (link below). This is an important one for me since I've personally spent more time as a last-mile analyst than anything else. That's why the last-mile will always be a first-class citizen in Zenlytic.

23

6 Kommentare

🔍 Breaking Down Snowflake's Earnings Call with ProntoNLP's AI-Powered Insights. Yesterday, Snowflake (SNOW) held its quarterly earnings call. The stock is down in pre-market by more than 10%. The company is still not profitable which might explain why it is down so much even though they did beat both on the top and the bottom line and raised guidence for product revenue. Using ProntoNLP's #GenerativeAI technology, we've analyzed the transcript to bring you key insights: 📊 Overall Sentiment: The LLM sentiment score slightly dipped from 0.88 to 0.84, but remains strong. 👥 Executive Confidence: Executives are overwhelmingly bullish, with 310 positive comments vs. 21 negative. CEO Sridhar Ramaswamy: 193 positive / 4 negative CFO Michael Scarpelli: 99 positive / 17 negative 📈 Analyst Perspective: Analysts also lean positive: 28 positive comments vs. 13 negative. 🔑 Key Aspects to Watch: "Forecast" and "Surprise" categories showed the most negative instances. 💼 Brokerage Firm Sentiment: Out of 12 firms on the call: 7 more positive than last quarter 4 more negative 1 unchanged Firms with improved sentiment: Deutsche Bank Citi Morgan Stanley Barclays Evercore Wells Fargo Truist Securities Firms with decreased sentiment: Wolfe Research, LLC Jefferies Needham & Company LLC UBS 📅 12-Quarter Trend: 28 LLM events declined vs. last quarter 21 improved Notable: "Forecast" category jumped from 0.32 to 0.84! 🤔 What This Means: Despite a slight overall sentiment dip, executives remain highly confident. The significant improvement in the "Forecast" category suggests optimism for future performance. The split in brokerage sentiment highlights diverse market perspectives. What are your thoughts on Snowflake's outlook? Let's discuss in the comments! #Snowflake #Quant #HedgeFunds #EarningsCall #FinancialInsights

4

🚀 #Microsoft releases SpreadsheetLLM, a model designed to optimize LLMs' powerful understanding on spreadsheets. One of the key advantages of SpreadsheetLLM is its ability to potentially query and manipulate spreadsheet data using plain English, rather than complex formulas or programming languages. This could democratize access to data insights and empower more individuals within an organization to make data-driven decisions. Furthermore, SpreadsheetLLM could help automate many of the tedious and time-consuming tasks associated with spreadsheet data analysis, such as data cleaning, formatting, and aggregation. #llm #generativeai #artifialintelligence

4

1 Kommentar

🔍 Just read a fascinating paper on the future of data ecosystem business models in the emergent "Artificial Intelligence of Thinks" (AIoT) space! Very useful systematic research that yields some key insights on how companies are transforming data into value in this developing ecosystem. https://lnkd.in/ert4nU7X 🔄 FOUR DISTINCT BUSINESS MODEL ARCHETYPES IN THE AGE OF AI ARISING 🤝 Adaptive Partnership - High control + High customization - Focuses on developing close, collaborative relationships with customers - Offers deep insights and personalized solutions - Best suited for strategic, long-term business relationships 👥 User-Centric - Low control + High customization - Prioritizes customer experience and flexibility - Enables broad market reach and scalability - Emphasizes user autonomy in configuring solutions ⚡ Standardized Efficiency - High control + Low customization - Focus on operational excellence and cost reduction - Streamlined processes and standardized offerings - Ideal for high-volume, routine operations [3] 🔄 Flexible Transaction - Low control + Low customization - Transaction-focused with minimal ongoing engagement - Emphasizes automation and self-service - Suitable for commoditized offerings 💎 AIOT BUSINESS MODEL VALUE CAPTURE STRATEGIES: 📊 Responsive Monitoring: Goes beyond traditional monitoring by enabling instant actions based on real-time data, including automatic corrective functions and predictive maintenance to prevent failures 🤖 Object Self-Service: Drives objects toward independent functionality with less human intervention, incorporating self-adjustment, replenishment, and self-operation capabilities 💡 Insights as a Service: A data-driven value proposition that leverages AIoT to collect, analyze and deliver real-time and collective insights that give meaning to massive amounts of data, enabling customers to make informed decisions   ✨ Enhanced User Experience: Focuses on improving customer interaction through digital add-ons and customizable interfaces and content that adapt to individual users' preferences and behaviors over time 💡💡 The research shows successful AIoT implementations balance control and customization to create sustainable competitive advantages. 💡💡 🤔 Where do you see your organization fitting in this framework? #AIoT #DigitalTransformation #Innovation #BusinessStrategy

15

1 Kommentar

I believe analytics databases / data warehouse and SQL can cover 80-90% of data use cases (structured + semi structured data). Even now, binary data can be processed using SQL (call external API to LLM). Take a peek at Muhammad Iqbal Effendy's piece to see how semi-structured data can be processed.

48

💸 Revenue Team Wants AI Costs, But Your MVP's Still Loading... 🤯 A Founding MLE Guide to Pre-MVP Cost Estimation 🔮 Initial Cost Estimation 1. Select Representative Models 📊:    Choose models at various scales. For NLP, consider BERT, GPT-J, (FLAN-)T5-XXL, and Falcon-40B. 2. Match Models to Hardware 🖥️:   Pair each model with appropriate GPU hardware. Example: GPT-J on A10G, Falcon-40B on A100 40GB. 3. Estimate Request Completion Time ⏱️:   Approximate time for each model to complete a request. Example:   - GPT-J: 1 second on A10G (made up number!)   - Falcon-40B: 10 seconds on A100 40GB (made up number!) 4. Calculate Hourly Costs 💸:   Research current GPU pricing. Example:   - A10G: ~$2 per hour (Modal Labs pricing)   - A100 40GB: ~$5 per hour (Modal Labs pricing) 5. Compute Cost per 1000 Requests 🧮:   Use the formula: (Seconds per request * Cost per hour) / (3600 seconds) * 1000   Examples:   - GPT-J: (1 * $2) / 3600 * 1000 ≈ $0.60 to serve 1000 requests   - Falcon-40B: (10 * $5) / 3600 * 1000 ≈ $3.00 to serve 1000 requests 6. Provide Order of Magnitude (OOM) Estimates 📏:   Present a range of costs based on different models. In this case, $0.60 to $3.00 per 1000 requests. 7. Factor in SLAs and Latency Requirements ⚡:   SLAs affect costs and can help constriant the solution space. For example, achieving a p99 latency of Xms might be 10x more expensive due to keeping a machine warm 🔧 Ongoing Cost Optimization (Thanks for the beautiful post Outerbounds) - Analyze Top-line Costs 📈:    Regularly review cloud bills to focus optimization efforts. - Identify Cost-driving Instances and Workloads 🔍:    Use tools to pinpoint expensive instances and tasks. - Monitor Resource Utilization 📊:    Avoid over-provisioning; pay attention to actual usage. - Optimize Workloads 🎛️:    Right-size resource requests based on real usage patterns. - Choose Optimal Execution Environments 🌐:    Leverage multi-cloud strategies for cost advantages. - Refine Based on Specific Needs 🎯:   Narrow estimates by understanding customer problems and required model scales. - Explore Serverless Options ☁️: - Stay Informed on Pricing 📚: Valuable resources: https://lnkd.in/gwragJBw https://lnkd.in/gVkGenhi https://lnkd.in/gq8pi4qd This framework is inspired by countless interactions with the ML/AI community #MachineLearning #CostEstimation #DataScience #AI #CloudOptimization

8

1 Kommentar

I used to think complex models were everything in AI. Then I learned: real value starts with simple, goal-aligned data. Simplicity drives results. 🧩 Want to hear more lessons from my data journey? Hit follow! #AIstrategy #DataDriven

5

1 Kommentar