Google announced Conversational Analytics in BigQuery via a Google Cloud blog post. The feature reframes analytics from writing SQL queries to asking questions in natural language, with BigQuery generating and running the underlying queries and returning results. It signals deeper LLM assistance inside the warehouse experience, likely including iterative refinement, explanation of outputs, and guardrails around data access and governance tied to BigQuery permissions.
Insights
Curated reads on Tableau, BI, analytics, and practical AI — short takeaways with links.
dbt Labs introduces ADE-bench, a benchmarking framework aimed at measuring how well AI agents and popular LLMs perform realistic analytics and data engineering tasks. It uses dbt projects to create reproducible, production like evaluation scenarios, likely including tasks such as writing SQL, updating models, fixing failing tests, and validating metrics. The announcement emphasizes standardized scoring and comparable runs across models and agent tools, making agent performance measurable beyond toy prompts.
Flerlage Twins published another installment of their Tableau tips series, explicitly noting the post includes many GIFs for step by step viewing. Expect a grab bag of practical techniques such as formatting shortcuts, calculations, parameter patterns, interactivity tricks, or performance friendly design moves that are easier to learn visually than from text. As part of a series, it likely links to earlier rounds and builds a reusable toolkit for daily Tableau work.
A Russian channel announces the release of two new chapters of an online book on building dashboards as a system, not a one off artifact. The linked page is Chapter 2, "Types of dashboards", which likely categorizes dashboard purposes and clarifies what a dashboard should be for in different contexts. It is positioned as structured guidance for BI practitioners, moving from visual output to repeatable processes and design decisions.
A Russian post announces a new initiative: an Open Data Visualization Academy presented by Alberto Cairo and collaborators. It promises many free lectures on data visualization, suggesting a structured learning hub rather than scattered talks. Although the post does not include the actual URL in the captured text, the announcement itself is meaningful for practitioners seeking foundational and advanced dataviz education from credible experts.
A short Russian Telegram post points to a Reddit thread claiming that Claude is now available inside Excel for Pro plan users. If accurate, this is a meaningful productivity shift: LLM assistance directly in spreadsheets where many analysts do cleanup, ad hoc modeling, and reporting. The linked discussion likely covers rollout scope, how to enable it, what tasks it supports such as formula help and text to table transforms, and any limitations.
dbt Labs explains what updates to the Open Semantic Interchange (OSI) spec mean for metric definitions, semantic portability, and AI. The post notes that an open OSI spec for semantic portability is now available and states dbt's commitment to operationalize semantics, signaling investment in standardized metric layers that can travel across tools. The theme is reducing vendor lock in and making consistent business logic accessible to BI and LLM assistants.
dbt Labs argues that inconsistent, duplicated metrics definitions, often called spaghetti metrics, will fail in AI driven analytics. The post advocates for a centralized metrics framework with consistent definitions, ownership, and governance so downstream consumers and AI assistants retrieve the same business meaning every time. It connects metric standardization to better semantic context, reduced ambiguity, and more reliable natural language analytics experiences.
dbt Labs publishes a deep dive on open table formats through a discussion with an Apache Iceberg and Apache Polaris contributor. It explains how Iceberg tables interact with the catalog layer, what responsibilities the catalog handles, and why catalogs matter for interoperability, governance, and cross engine access. The post is positioned as a comprehensive explainer for teams building lakehouse architectures and evaluating open standards versus vendor specific stacks.
A Russian dataviz post revisits a confusing RBC infographic about snow level measurement and explains why the chosen reference point made the chart misleading. The author links back to an older critique and likely contrasts proper measurement baselines, units, and annotation choices that prevent viewers from inferring nonexistent extremes. The core is a practical lesson: define the metric and reference point before plotting, or you will manufacture drama.
Duplicate of the snowfall measurement critique posted in the same channel. It references an older Telegram post that complained about an unclear RBC infographic and uses that case to discuss how not to measure or present snow levels. Expect guidance on choosing a consistent baseline, clarifying what the metric represents, and avoiding visual storytelling tricks that come from arbitrary reference points rather than real data behavior.
A Russian post from @data_bar narrates how a side project evolved into a real data product built by a group of four. It reads like a behind the scenes case study, including team roles and decisions that turned an idea into something shippable. The post links to a teammate profile or channel, suggesting additional context about the build process, tooling, or product framing. More product and execution oriented than pure AI, despite the channel topic tag.
The @data_csv Telegram channel announces enrollment for a three month data visualization course titled “Aesthetics in Charts,” promoted as the author’s flagship program and opened only a few times per year. The post emphasizes a limited enrollment window and maximum discounts, positioning it as a structured training path rather than a single workshop. No syllabus, tools, or projects are included in the snippet, so specifics are unclear.
The @datavizcomics Telegram channel promotes access to a recorded workshop called “Music Poster,” where participants created posters and videos inspired by favorite songs. The post offers the recording and all materials for a fee, suggesting a creative data viz or generative design session rather than BI dashboarding. No details are provided on tooling, data sources, or techniques used to produce the posters and video outputs.
A Russian channel summarizes a New York Times interactive landing page that alleges corruption by arguing Trump and his family earned large sums by leveraging presidential status. The post positions it as a compelling interactive narrative with quantified claims and likely supporting charts, timelines, or networks. It is more journalistic dataviz than BI practice, but it can be useful as a reference for structuring evidence, sourcing, and interactive explanations.
A Russian post discusses the trend of AI systems that can act in the real world, for example browsing the web and clicking UI, and introduces OpenClaw alongside the idea of building a social network for bots. The framing is about turning agent capabilities into coordinated, tool using entities rather than isolated chatbots. While not a BI tutorial, it signals where automation is heading: agent to agent interaction, shared environments, and new surfaces for monitoring and governance.
The @datavizcomics channel recommends a book sold via an Ozon product link, describing it as a travel story where charts, diagrams, and tables are integrated into the narrative rather than presented as boring appendices. The post suggests a hybrid of fiction and data visualization, aimed at making quantitative artifacts emotionally engaging. No author, sample pages, or concrete visualization techniques are included beyond the concept and purchase link.
A Russian language post recommending a difficult, emotionally heavy visualization project by Nir Smilga, described as one of the author’s favorite creators. The project focuses on abducted and murdered people from Smilga’s homeland and is shared to raise awareness and prompt reflection. The exported item does not include the actual project link or details on the tooling, but it is positioned as a noteworthy visualization piece worth seeing despite the painful topic.
A Russian post argues that the global AI boom is not only software, it is physical infrastructure: data centers, power generation, and networks. It frames a growing risk of energy scarcity for compute and links it to oil and gas, utilities, and social platforms competing for capacity. While not a how to guide, it is a useful macro lens for analytics leaders planning AI roadmaps, cost models, and sustainability reporting.
A Russian post shares a quick review of the book "Dashboards that Deliver" and highlights a section on defining and truly understanding the user throughout the dashboard process. The linked site likely describes the book, author, and framework for going from requirements to usable dashboard decisions. The key takeaway is product thinking applied to BI: audience, context, and decisions drive design more than chart cleverness.
A Russian post announces the start of a live dashboard intensive covering dashboards from A to Z, with a small group, chat, and feedback. It suggests there is a structured program shown in a picture in the original post, but the captured text does not include the curriculum details or tooling. As a training offer, the value depends on the instructor credibility and the specificity of topics like requirements, layout, interactivity, and delivery.
A Russian post shares an interactive visualization showing when the last train departs across different Tokyo rail lines and stations. The author calls it impressive and sticky, while admitting it is not very useful in practice. The linked site likely includes an interactive map with time encoded styling and filtering by lines or stations. It is primarily inspiration for temporal mapping and interaction design rather than a BI method.
A Russian post bluntly states that data governance is exhausting and argues the pain is justified because teams lack strong examples of measuring governance impact on the business and ways to sell it internally. It mentions typical roles like data owners and data stewards and implies common failure modes: bureaucracy without value proof. While informal, it surfaces a real analytics management gap: governance KPIs and outcome based framing.
A Russian language post from @leftjoin titled "Bluff, gaslighting and betrayal among AI" starts by describing the simple but relationship destroying game So Long Sucker, comparing it to card games, Uno, and Monopoly. The provided excerpt suggests the author uses the game’s incentives and coalition dynamics as a lens to discuss deceptive behavior and manipulation in AI systems or AI discourse, but the text is truncated and no link is included.
A Russian post announces an open issue of the newsletter "This Is Not an Indicator" aimed at helping readers better understand Russian official statistics. It mentions that since launching in February 2025, they have compiled source maps on topics like disability and other social measures, suggesting a methodological focus on data sources, definitions, and common misreadings. It is more statistical literacy than BI tooling, but relevant for analysts working with official data.
A Russian post notes that HeadHunter released a report on the future of skills centricity, arguing the labor market is shifting away from job titles and diplomas toward measurable skills. The PDF is said to be in the comments, which likely includes frameworks for skills taxonomy and assessment. For analytics leaders, it connects to how to design career ladders, training plans, and hiring rubrics for data roles based on capabilities rather than labels.
FlowingData published its monthly roundup of visualization tools and resources for January 2026. These roundups typically compile new releases, libraries, tutorials, and noteworthy projects across the dataviz ecosystem. While not specific to one BI platform, it can be a useful scanning list for practitioners who want to keep up with charting tools, mapping resources, and workflow utilities without tracking every source individually.
The @datavizcomics channel notes that the media project “If It Were Accurate” has introduced a new content format, described as delightful and including small data challenges. The post is a recommendation rather than a detailed breakdown, and no link is provided in the snippet to see the new format directly. The value likely comes from short exercises that encourage readers to interpret charts or work through small analytical tasks.
dbt Labs announces the call for proposals for dbt Summit and outlines what makes a strong session: real problems, concrete solutions, and practical lessons learned from shipping analytics work. It positions the event as a venue for case studies and implementation details, and includes a submission deadline of March 31. Useful for teams considering speaking about governance, modeling, testing, or production rollouts.
FlowingData flags a policy change to the US Affordable Care Act subsidy scale that took effect in January and effectively reintroduced a sharp eligibility cliff. The post points to how small income changes can now trigger large swings in premiums and out of pocket costs, creating discontinuities that matter for analysis, forecasting, and equity discussions. It frames the cliff as a measurement and communication problem, not just politics.
Mozilla is adding explicit controls in an upcoming Firefox release that let users turn off browser AI features. The change signals a shift toward opt out and transparency for AI that can touch browsing behavior, search, and assistive experiences. For analytics teams, it is a reminder that AI driven UI changes and privacy settings can alter consent rates, tracking coverage, and user flows across web properties.
FlowingData highlights New York Times reporting by Ben Casselman: a US government shutdown is delaying Bureau of Labor Statistics releases, including key jobs reports that many teams use for economic monitoring. With BLS staff and publication workflows disrupted, scheduled labor market updates may slip, creating temporary gaps in official time series and uncertainty for analysts tracking employment trends.
FlowingData highlights a proposal framed as cost saving: shifting parts of decennial census counting to US postal workers. The post points to the operational and statistical implications of changing collection methods, such as coverage error, inconsistent protocols, nonresponse handling, and comparability with prior censuses. It is a reminder that data pipeline changes at collection time can dominate downstream analytics and trend interpretation.
FlowingData shares “Isometric NYC” by Andy Coenen, a pixel art, isometric rendering of New York City that sits between curiosity and city visualization. The emphasis appears to be on artistic representation and the craft of building an isometric view rather than analytical insight, modeling, or BI workflows. As a reference, it can inspire map styling and perspective choices but does not provide a method tutorial in the snippet.
FlowingData notes that US measles was declared eliminated in 2000, but cumulative case counts are rising again, tied to a South Carolina outbreak and broader vaccination context. The post likely critiques how cumulative charts frame public health risk, emphasizing temporal accumulation, seasonality, and the communication tradeoffs versus incident counts. It is more about data storytelling and public health monitoring than BI tooling.
FlowingData highlights a New York Times visualization approach for showing shifts in net support for the president, focusing on “U turns” in voter sentiment. The emphasis is likely on a chart form that makes reversals visible, such as slope changes, net support lines, or segmented trend comparisons. While useful as a communication pattern, the snippet does not indicate a replicable tutorial or the underlying data processing steps.
FlowingData highlights the newest US Census Bureau population estimates, showing which states gained residents and which saw net out migration. The piece focuses on the pattern across all states, emphasizing that most grew while a smaller set declined. It is primarily a data story and visualization oriented roundup rather than a methods tutorial, but it can be useful as a reference dataset for mapping and trend analysis examples.
FlowingData flags reporting on Roblox using AI based age verification and argues it is a weak approach for a platform heavily used by children. The core issue is model reliability and the risk of false accepts or rejects when identity or age is inferred from images or signals. For analytics and data teams, the post is a cautionary example of deploying probabilistic ML for high stakes gating, where error rates translate into safety and compliance exposure.
Storytelling with Data shares a practical redesign approach for bullet heavy slides and dashboards that overwhelm readers. The post focuses on turning dense, multi level bullet lists into scannable communication using intentional hierarchy, whitespace, grouping, and emphasis. It likely demonstrates before and after examples, shows how to keep one idea per line, and recommends alternatives like short labeled sections, callouts, or visual structure instead of long bullets.
FlowingData points to IsoCity, an open source city building game inspired by SimCity. While not BI focused, it is a simulation project that can expose interesting modeling concepts like resource constraints, growth dynamics, and system feedback loops. For analysts, it is mainly a curiosity and potential sandbox for experimenting with telemetry, event modeling, and scenario analysis if you instrument gameplay data.
GraphQL for Advanced Tableau Metadata Analysis
Master Tableau's GraphQL Metadata API for advanced lineage analysis, impact assessment, and automated data catalog creation. Includes practical queries and real-world examples.
Deep Dive into Tableau PostgreSQL Repository
Master Tableau Server's PostgreSQL repository for advanced metadata analysis. Learn SQL queries for permissions matrix, user management, and dashboard usage analytics.
Mastering Tableau REST API for BI Automation
Complete guide to Tableau REST API automation. Learn authentication methods, practical Python examples, and real-world automation use cases for Tableau Server management.
Tableau Cloud vs On-Premise: Metadata Access Comparison
Complete comparison of metadata access capabilities between Tableau Cloud and on-premise installations. Learn about PostgreSQL repository, REST API, and GraphQL differences.
Why Metadata Matters in BI Product Development
Transform your BI approach by treating Tableau Server as a product. Learn why metadata matters for user activity tracking, data governance, and performance optimization.
Simple Yet Effective: Practical Tableau Tips
Essential Tableau tips and tricks for more efficient dashboard development and better visualizations. Practical advice from real-world experience.