Non-technical people are now generating massive amounts of data code through AI agents instead of waiting for BI teams, but this code lives everywhere—scattered across chats, laptops, and Slack—with no way to verify accuracy, reproduce results, or enforce standards. Traditional BI tools can't fix this because they're too rigid, leaving data teams stuck between chaos and lockdown.

• Millions of non-technical users are writing billions of lines of AI-generated code for analytics, replacing the old ticket-and-wait model with instant answers—and nobody wants to go back.
• The generated code is ungoverned and untraceable: there's no record of what context the AI used, which tables it queried, or what filters it dropped, making it impossible to verify if the numbers are actually correct.
• Data teams face a false choice: either let people generate whatever they want and abandon governance, or force them back into rigid BI tools that can't do much of anything.

DuckLabs, the company behind the popular DuckDB analytics database, is being acquired by AWS in early September. The core open-source projects will remain free under the MIT license and governed by the nonprofit DuckDB Foundation, but the team gains AWS's infrastructure and reach to scale the technology further.

• DuckDB hits over 1 million downloads daily and the founders realized their bootstrapped model couldn't support the project's growth trajectory without becoming a bottleneck
• All core Duck Stack components (DuckDB, DuckLake, Quack) stay open-source under MIT license with the DuckDB Foundation maintaining stewardship
• AWS has committed long-term support and the DuckLabs team of 30+ people will remain together in Amsterdam, focusing on technical work rather than sales operations

Claire Gouze lays out a practical process to build and test a context layer your data agents can trust. She shows how to start small with dbt docs and markdown files, write evals, clean up data models, and iterate until you reach around 90% reliability.

• Adding query logs and profiling alone got the agent stuck around 40% accuracy; reworking the data model and documentation pushed it to 90%.
• Treat context like code: store it in files/markdown, version in Git, add CI/CD evals, and iterate on failure cases rather than plugging agents directly into raw databases.
• Start small—10 to 20 core tables with existing dbt docs—rather than trying to cover the whole warehouse at once.
• Evals should come from real BI query history or autogenerated question sets, reviewed by teams and tracked over time to build trust in the agent.

The article argues that technically sound data teams often stop at delivering dashboards and pipelines, failing to influence actual decisions. It introduces a three-layer Data-Perspective-Action framework and practical habits—like weekly one-pagers—to build the interpretive “Perspective” layer that links data work to business outcomes.

• One audit found stakeholders opened only 10 of 200 working dashboards before making decisions—the rest were accurate but ignored
• 93% of leaders blame culture and change management, not technology, as the top barrier to data-driven decisions (Bean's AI & Data Leadership survey)
• As AI automates pipeline/dashboard work, the "Perspective" layer—adding context and recommending action—becomes the main source of a data team's irreplaceable value
• Teams that skip interpretation and just hand raw analysis to busy executives are the ones most likely to become redundant

GitHub built Qubot, an internal Copilot-powered analytics agent that lets employees query their data warehouse in plain English via Slack, VS Code, or the CLI. It pulls federated context from raw, conformed, and curated data layers, then runs queries on Kusto or Trino to deliver SQL-backed insights within seconds. A continuous evaluation framework and community contributions to the context layer keep responses accurate and fast.

• Qubot lets any employee query GitHub's data warehouse in plain English via Slack, VS Code, or CLI, cutting analytics team question load and freeing analysts for harder work
• It uses a three-tier federated context layer (bronze/silver/gold) maintained collaboratively by product, analytics, and business teams, turning scattered knowledge into a single source of truth
• Investing in well-structured context both improved accuracy and cut answer time by two-thirds
• A continuous offline evaluation framework tests every context or agent update against known SQL answers to catch regressions before release

This policy explains what personal data a16z collects when you use the Speedrun site and related services, where it comes from, how cookies and analytics track your activity, and how your information is used, shared, and protected. It also covers your rights under laws like the CCPA and how to manage or opt out of data collection.

• a16z's Speedrun collects broad personal data (contact info, IP, demographics, education/work history) directly from users and from third-party vendors, analytics, and ad partners
• Cookies on the site range from essential login functions to performance/ad-tracking tags, and users can opt out via browser settings or Google Analytics opt-outs, though this may break some features
• Collected data is used not just for site operation but for marketing, fraud detection, career counseling, office access, and combining with outside data sources for personalization
• The policy binds anyone interacting with Speedrun (including portfolio-company reps, partners, vendors) but excludes non-a16z-controlled firms and job applicants/hires

mviz is a Claude skill that simplifies the creation of static reports for ad hoc data analysis by converting compact JSON specifications into professional HTML visualizations. It emphasizes a fast, AI-driven workflow that allows users to iterate quickly, generate reports, and utilize a variety of chart types without extensive coding. The tool works seamlessly with data from various sources, including local files and cloud databases.

• Mviz is a Claude skill that turns compact JSON specs into polished HTML reports/charts, prioritizing speed over BI-style reusability.
• Workflow: build context via natural-language queries (Claude writes the SQL), form a narrative, generate an initial chart, then iterate with follow-up commands.
• Supports many chart types (bar, line, heatmaps, funnels, tables, KPIs) laid out in a customizable 16-column grid.
• Can pull data directly from local files or cloud databases, producing clean, responsive output suitable for print or digital sharing.

Data trust is not achieved through a single tool but rather through addressing numerous small failures throughout the data lifecycle. The article illustrates this concept using a real-life incident of clickstream data discrepancies, highlighting the importance of rigorous practices in instrumentation, testing, and pipeline management to prevent data trust erosion.

• A real clickstream data discrepancy incident traced back to small failures across the pipeline, not one big bug, illustrates how data trust actually erodes.
• Preventing this requires rigor at multiple stages simultaneously—instrumentation, testing, and pipeline management—since fixing just one won't restore trust.
• Data trust is framed as "death by a thousand paper cuts": it's lost gradually through many minor issues rather than a single catastrophic failure.

Livedocs is a collaborative platform that merges the functionality of notebooks with app-building simplicity, ideal for various data tasks such as exploration, analysis, and visualization. It supports powerful AI tools, enabling users to perform advanced analytics, create interactive dashboards, and share insights effortlessly.

• Livedocs combines notebook-style flexibility with app-building simplicity for data exploration, analysis, and visualization
• Built-in AI tools enable advanced analytics and let users build interactive dashboards without heavy manual coding
• The platform emphasizes collaboration, making it easy to share insights and outputs with others