What is one specific practice or element that is essential for building a reliable semantic layer to support AI-driven analytics in organizations, and why?

Most answers emphasize the need for trusted, well-governed, and clearly defined data and metadata as foundational to building a reliable semantic layer for AI-driven analytics.

12 answers

Answers

  • Matt Ausman

    Chief Information Officer · Zebra Technologies

    Data Governance

    Matt Ausman says data governance is essential, pointing to Zebra's federated model with data owners per function, because running AI on unvalidated data or shifting definitions will give bad results.

    The answer starts at 16:17.

    Technovation with Peter High (CIO, CTO, CDO, CXO Interviews) · Listen on Apple Podcasts
  • Shayde Christian

    Chief Data & Analytics Officer · Cloudera

    Technical and Business Metadata

    Shayde Christian says technical and business metadata was very important in creating a semantic layer, and that his organization made a lot of strides in that area while continuing to build toward a true semantic layer.

    The answer starts at 31:07.

    Chain of Thought | AI Agents, Infrastructure & Engineering · Listen on Apple Podcasts
  • Mike Aronson

    Vice President, Global Talent Acquisition · Johnson Controls

    Central Ontology

    Mike Aronson says the ontology part is where defining information once and reusing it across models lives, and describes comparing semantic views to find common pieces across multiple models.

    The answer starts at 56:43.

    Explicit Measures Podcast · Listen on Apple Podcasts
  • Martin Casado

    General Partner · Andreessen Horowitz

    Open Table Formats

    The passage does not show Martin Casado giving this answer; the recommendation to use open table formats, specifically Iceberg, is made by George Fraser.

    The Analytics Engineering Roundup · The Analytics Engineering Roundup · Read
  • Anahita Tafvizi

    Chief Data & AI Officer · Snowflake

    Unified Metric Definitions

    Anahita Tafvizi says organizations need unified definitions for their metrics, because without that foundation AI can just scale your chaos across the company.

    New York Stock Exchange · Watch on YouTube
  • Jess Ramos⚡️

    Founder & Content Creator (Tech, Data, & AI) · Big Data Energy Analytics⚡️

    Trusted Data

    Jess Ramos says having the right trusted data or trusted intelligence is going to be critical for AI to work, because an AI agent given the wrong information will fail, hallucinate and do the wrong thing.

    Jess Ramos · Watch on YouTube
  • Mike Ferguson

    Owner, CEO and Head of Research · Intelligent Business Strategies

    Data Concept Models

    Mike Ferguson says he is seeing organizations go back to their roots by getting data concept models in place, then building and curating data around them to achieve a systematic, repeatable approach.

    The answer starts at 12:57.

    It's About Data · Listen on Apple Podcasts
  • Francis Rose

    Host & Co-founder · Fed Gov Today

    Understanding Data Landscape

    Francis Rose says the most important thing is understanding the landscape of your data, including its full scope, where it comes from, and who should access it, because the models and AI tools are only as good as that data.

    Fed Gov Today · Watch on YouTube
  • Gaurav Pathak

    Senior Vice President Product Management · Salesforce

    Single Source of Truth

    Gaurav Pathak says organizations need a single source of truth for metadata and data, framing it as essential for getting agentic AI right and worth a big investment in data.

    Super Data Science: ML & AI Podcast with Jon Krohn · Watch on YouTube
  • Tristan Handy

    CEO & Founder · Dbt Labs

    Clean Gold Layer

    Tristan Handy says a clean gold layer is essential, meaning high-quality data with no duplication of tables or columns, and eliminating that ambiguity is really important because agentic analytics plugs directly into the gold layer.

    The answer starts at 19:15.

    DataFramed · Listen on Apple Podcasts
  • Sam Pierson

    Chief Technology Officer · Qlik

    Quantifiable Data Reliability

    Sam Pierson says having a data foundation that is quantifiably reliable and high quality is the first thing his team found necessary, along with adding context on top of the data to improve the accuracy of LLM answers.

    Data Faces Podcast · Data Faces Podcast · Listen at 8:12
  • Muralidhar Vemulapalli

    Chief Enterprise Architect/ Acting CTO · Arkansas Blue Cross And Blue Shield

    Enterprise Semantic Layer

    Muralidhar Vemulapalli says defining an enterprise semantic layer is the most important element, because machines, unlike developers, do not know what column names mean.

    CDO Magazine Podcast Series · CDO Magazine Podcast Series · Listen at 4:11

More Data Management questions

View all →