George Williams

Please stop calling every pipeline a “data product.” I see it a lot, teams proudly calling their dashboards or ETL pipelines “data products.” But a real data product is way more than just moving or storing data. It’s something people can trust, use, and actually make decisions with. It has clear ownership, context, and guarantees for the businesses relying on it. From my experience running enterprise analytics projects, the difference between a pipeline and a true data product is night and day. Here’s what sets it apart: - Purpose drives design: A true data product solves a specific business problem, not just generates numbers. - Context is key: Metadata, lineage, and business meaning make data understandable and actionable. - Accessibility is non-negotiable: It should be discoverable, self-service, and seamlessly integrated into workflows. - Operational guarantees: Reliability, freshness, and consistent behavior are what separate a product from a pipeline. Pipelines are the engine, but the data product is the experience for the user. Without the human-led strategy, governance, and adoption focus, AI or automation alone cannot turn raw data into decision-making power. When was the last time your team built a “data product” that people actually relied on, and not just admired on a slide?

117

28 Comments

Sometimes when I'm waiting for a SQL query to finish running, I'll randomly slack someone and ask "are you joining this meeting?" and then go silent. Good way to keep people on their toes.

86

12 Comments

Dynamic Visibility Control in Tableau is a tool you can use even without any RLS. It’s usually for things like drill-downs, like when you choose a country and a city chart appears, or just a simple "show/hide details" switcher. Hovewer for RLS, it solves a specific problem: Blank Spaces. Read more about it in my blog: https://lnkd.in/dTpnEAdX #Tableau #RLS #DynamicZoneVisibility #DataViz

3

You trained the model. You trusted the output. But can you trace the data that led to the decision? 👀 AI in analytics isn’t dangerous because it’s wrong. It’s dangerous because it’s convincing, even when it’s wrong. 📰 This article breaks down the real risks of AI-powered insights and how to design for trust with traceability, governance, and human oversight. Read it here: https://lnkd.in/eKuQnv7J

10

Migrating to Snowflake? Don’t just move faster. Move smarter. 👇    Too many migrations copy yesterday’s complexity into tomorrow’s platform. Legacy reports. Duplicated KPIs. Unused dashboards. Technical debt — just relocated.    At BlueCloud, we’re seeing a different pattern emerge: AI-guided migrations that clean, rationalize, and modernize before scale.    Here’s what that looks like in practice:    ➡️ Rationalize your BI landscape before migration  ➡️ Align metrics and eliminate redundant logic  ➡️ Prioritize key data sources   ➡️ Build a phased roadmap tied to measurable outcomes    For executives, this means lower cost, clearer governance, faster time to value, and a foundation designed for AI from day one.    Learn more: https://lnkd.in/dGdyXx-s    #Snowflake #CloudMigration #EnterpriseAI #DataStrategy #ModernData 

26

Your best data people didn't sign up to be firefighters.  Every month, it’s the same scramble.  Patch the pipelines. Clean the spreadsheets. Fix the broken dashboards.  The more manual your fixes, the less strategic your team becomes.  Thats not innovation. Thats survival.  We break down the real cost of reactive data management – and how to escape the cycle.  

5

1 Comment

One thing I keep seeing in enterprise AI rollouts: the highest-risk data usually isn’t in tables it’s in the messy stuff. Tickets, call transcripts, notes, emails… the place where PII shows up casually. That’s why I liked seeing Snowflake move in two practical directions recently: A clearer path to structuring unstructured data with Cortex AI Functions, so teams can extract consistent signals without building a one-off pipeline every time. AI_REDACT reaching GA automated PII detection + redaction for unstructured text, built into the platform. From an AI & Security Platform lens, this is the kind of “boring guardrail” that actually enables speed: you reduce privacy risk upfront, and you make cross-team alignment (engineering / data / security) much less painful. If you’re working with unstructured data today, where do you enforce redaction at ingestion, at query-time, or right before model access?

1

Databricks just launched Genie Code and I'm really excited about this one!                           It's not another code completion tool, it's an autonomous AI agent that can reason through problems, plan multistep approaches, and execute full data and AI workflows end to end. The real superpower here? Unity Catalog. Having an AI coding agent that actually understands your data, the schemas, lineage, governance, business semantics, is what really separates this from other coding agents when it comes to complex data tasks. It's not just writing code in a vacuum, it knows your data. Talking with our Oil & Gas customers, I've seen firsthand how much this can seriously expedite data and ML workflows. Teams in this space are constantly stretched thin and dealing with massive, complex datasets. Having an AI partner that can actually plan and execute, not just suggest snippets, is a huge unlock.

13

AI can write the SQL. But SQL against what definitions? This week's Data Report: Joe Reis says the reckoning is here, and he's right about the capability. But every other story I tracked points at the same gap. A pipeline ran green and delivered zero rows. 50+ data quality tools exist and most teams use none. Three conversations landed on ontology as the missing layer, independently. The reckoning isn't about whether AI can write the code. It's about whether your org has defined what "correct" means. https://lnkd.in/gENSNk_T

1