Forget AGI. Your AI Is Dumb Without Your Data | Josh Howard, Databricks

Forget AGI. Your AI Is Dumb Without Your Data | Josh Howard, Databricks

"Without context, your agents are dumb." That's how Josh Howard, Senior Director of Product Marketing for Executive Audiences at Databricks, closed Episode 40 of the Data Faces Podcast.Frontier models are some of the most advanced technology of our lifetime. They are also dumb in the way that matters for your business, because they were trained on the public internet and have never seen your customer records, your forecast methodology, or your sales policies. In this episode, host David Sweenor and Josh Howard unpack new findings from the Databricks and Economist Enterprise *Making AI Deliver* survey of 1,221 senior technology leaders, including the 84/43 measurement gap, why data infrastructure costs more than the GPU bill, where AI agents are already working in the enterprise, and why the real race over the next five years isn't to AGI.**Key Takeaways:**1. Today's models are "dumb" not because they lack capability, but because they lack enterprise context2. 59% of senior tech leaders say data storage and movement is the biggest AI cost, only 25% say compute3. 84% of executives say AI is beating expectations, but only 43% require teams to measure the impact4. AI agents now create 80% of new databases on Databricks' Neon serverless Postgres layer, up from 0.1% in 20235. The next five years will reward boring work: cleaning data, fixing semantics, and tying agent projects to measurable outcomes**Timestamps:**00:00 - Opening and introduction01:17 - Josh's role leading PMM for executive audiences at Databricks02:21 - If not PMM: full-time fly-fishing guide in Colorado03:23 - "Your AI is dumb" — what the phrase actually means05:25 - Structured vs. unstructured data and the industry's row-and-column trap06:20 - Where Josh and Dave first met at Dell Technologies08:13 - Metadata, context, and the 20-year-old enterprise architect fight09:37 - The November 2022 ChatGPT moment in the C-suite11:07 - Trying to pry Excel from a financial analyst's hands at Alteryx12:08 - Human-in-the-loop and the Replit agent that wiped a production database12:53 - Conversational analytics, Databricks Genie, and internal semantics19:11 - Inside the Databricks and Economist *Making AI Deliver* survey20:54 - The 84/43 measurement gap23:21 - The 59/25 cost split — data infrastructure vs. compute28:30 - Upskilling, prompt engineer hype, and behavior change30:17 - AI washing on the 101 corridor and Allbirds' pivot to NewBird AI33:26 - What will look obvious in 202735:39 - Closing thought: "Without context, your agents are dumb"**More insights and resources:**Blog: https://tinytechguides.com/blog/forget-agi-your-ai-is-dumb-without-your-data/?utm_source=youtube&utm_medium=video&utm_campaign=ep40-josh-howard&utm_content=description Survey: https://www.databricks.com/resources/analyst-research/making-ai-deliver**Connect with Josh Howard:**LinkedIn: https://www.linkedin.com/in/joshoward/Databricks: https://www.databricks.com/Drop your thoughts in the comments!Like, share, and subscribe for more insights.#AgenticAI #EnterpriseAI #DataLeadership #Databricks #DataFacesPodcast

Metadata, semantics, and the future of AI context | Steve Wooledge

The difference between an AI that "hallucinates" and one that acts intelligently lies in context. In Episode 39 of the Data Faces Podcast, Steve Wooledge (CMO at Collate) joins David Sweenor to discuss why metadata, once a technical "card catalog," is now the foundational layer for the agentic era. Steve traces his journey from chemical engineering to building categories at Alteryx and Alation, and now leads the charge for open-source semantic intelligence at Collate. Key takeaways: 1. Metadata vs. semantics. Technical descriptions aren't enough for AI, and Semantic Intelligence Graphs provide the "gut feel" AI lacks. 2. The Switzerland approach. Organizations need a neutral metadata layer that spans silos such as Databricks and Snowflake. 3. Marketing velocity. AI is compressing production workflows and "Taste Squared" is the new metric for human marketing leaders. 4. The category creation playbook. Steve shares lessons learned from defining "Agentic Data Intelligence" at Alation. Chapters ● 0:00 – Introduction ● 1:08 – From Chemical Engineering to Data Sales ● 3:45 – Guitar Shredding and "Melodic" Hard Rock ● 5:01 – Marketing Lessons. Dave Kellogg and the power of first principles ● 7:40 – The hard truth about partner marketing and global SIs ● 10:18 – Why open source out-innovates the enterprise ● 15:56 – Metadata for AI agents. The semantic intelligence shift ● 20:45 – The "neutral layer" strategy ● 24:03 – How AI is changing the CMO role ● 27:23 – "Taste squared." Why you can't be a lazy marketer ● 32:19 – Career advice for the next generation of data professionals ● 36:28 – Final advice. Peer review and quality controlConnect with Steve ● LinkedIn: https://www.linkedin.com/in/stevewooledge/ ● Collate: https://getcollate.io/Follow TinyTechGuides -- Blog: https://tinytechguides.com/blog/why-ai-agents-require-a-switzerland-approach-to-metadata/?utm_source=spotify&utm_medium=video&utm_campaign=ep39-steve-wooledge&utm_content=description --Substack: https://open.substack.com/pub/davidsweenor/p/why-ai-agents-require-a-switzerland?r=1s6e48&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true #datafacespodcast #AI #Metadata #DataGovernance #B2BMarketing

Bots Need Not Apply: Authentic Voices in Data and AI | Kate Strachnyi

LinkedIn has a "rewrite with AI" button. Meanwhile, Kate Strachnyi is building an entire media company on authentic human voices. Is she right?In Episode 38 of the Data Faces Podcast, Kate Strachnyi (Founder, DATAcated) shares how she pivoted from finance to data visualization, built a 40+ creator influencer agency, and why she's betting on real humans over AI-generated content.Key Takeaways:1- How Kate followed the revenue data from courses and books to a focused media business2- The DATAcated Plus model: matching authentic creators to brand campaigns in data and AI3- Why Kate calls AI-rewritten content "non-GMO" and holds her creators to the same standard4- The shift from "Kate = DATAcated" to an agency brand that scales beyond one person5- The 20-year question: who fact-checks AI when today's subject matter experts retire?Timestamps:00:00 - Opening0:05 - Kate's background and what DATAcated does2:10 - Pre-finance Kate: what she wanted to be before data found her3:05 - The career pivot from risk management to data visualization5:03 - How DATAcated evolved from training to a media company7:27 - How the influencer model works behind the scenes9:33 - Automating business operations with Claude Code11:01 - Walking the line between brand amplification and spam14:11 - The fake tattoo story from Big Data London15:03 - DATAcated Plus vs. analyst firm engagements17:14 - Sold-out personal branding session at Gartner with Scott Taylor22:15 - Shifting from "Kate = DATAcated" to an agency brand24:02 - What works on LinkedIn now vs. five years ago27:01 - AI-generated content and the "non-GMO" philosophy29:04 - The 20-year question: who fact-checks AI when the experts retire?30:20 - Deep fake Dave and why Kate plans to remain authentic31:24 - Betting on AI for operations while keeping creative output human33:57 - Does AI make you more productive or just busier?36:19 - Where to find Kate and DATAcatedMore insights and resources:Blog: https://tinytechguides.com/blog/Connect with Kate Strachnyi:LinkedIn: https://www.linkedin.com/in/kate-strachnyi-data/DATAcated: https://datacated.com/Drop your thoughts in the comments!Like, share, and subscribe for more data and AI conversations.#AuthenticContent #AI #DataFacesPodcast

Why Bad Data Didn't Matter Until Now | Brendan Grady

For 25 years, data quality was everyone'sproblem and nobody's priority. Brendan Grady, EVP and GM of Analytics & AIat Qlik, explains why the stakes just changed. In this episode recorded on location at QlikConnect 2026, David Sweenor and Brendan discuss consequence management, whereenterprise agentic adoption really stands ("prior to stage zero"),Qlik's Trust Score for AI, the shift from dashboards to decision intelligence,and why open standards like MCP matter in an agentic world. For 25 years, data quality was everyone's problem and nobody's priority. Brendan Grady, EVP and GM of Analytics & AI at Qlik, explains why the stakes just changed. In this episode recorded on location at Qlik Connect 2026, David Sweenor and Brendan discuss consequence management, where enterprise agentic adoption really stands ("prior to stage zero"), Qlik's Trust Score for AI, the shift from dashboards to decision intelligence, and why open standards like MCP matter in an agentic world. Key takeaways: 1. Data quality was never fixed because there were no consequences for getting it wrong. AI agents changed that equation. 2. Enterprise agentic adoption is in its earliest days. Customers are experimenting, but production-grade agents are rare. 3. Qlik's Trust Score for AI gives decision-makers a quantifiable measure of data quality before it reaches an agent. 4. "Dashboards are dead" as a destination, but the data and decisions they inform are more important than ever. 5. Data professionals should become data product owners and trusted guides as agents take on routine work. Chapters: 0:00 Introduction at Qlik Connect 2026 1:14 Brendan's first job: Sound of Music tourguide 2:04 Lessons from the early analytics era 3:32 Why data quality has never been fixed 4:46 Consequence management in the agentic era 6:08 Where agentic adoption actually stands 7:46 Future-proofing against LLM shifts 8:24 The analytics engine and unknown unknowns 10:29 Structured vs. unstructured data 12:04 Hallucinations and trust scores 15:30 "Dashboards are dead" 18:05 Brain outsourcing and cognitive debt 21:57 MCP server and open standards 23:54 Qlik 2026 themes: trust, context,flexibility 26:12 Advice for data professionals 28:15 Does AI expand who can participate inanalytics? Links: Blog post: https://tinytechguides.com/blog/why-bad-data-didnt-matter-until-now/ [https://tinytechguides.com/blog/why-bad-data-didnt-matter-until-now/] BrendanGrady on LinkedIn: https://www.linkedin.com/in/brgrady/ [https://www.linkedin.com/in/brgrady/] Qlik: https://www.qlik.com/ [https://www.qlik.com/] Data Faces Podcast: https://tinytechguides.com/data-faces-podcast/ [https://tinytechguides.com/data-faces-podcast/] Subscribe: https://www.youtube.com/playlist?list=PLzrDACjTQ4OBoQ8qM1FMGBwYdxvw9BurR [https://www.youtube.com/playlist?list=PLzrDACjTQ4OBoQ8qM1FMGBwYdxvw9BurR] #DataFacesPodcast #QlikConnect #AgenticAI#DataQuality #DecisionIntelligence

A-Eye Gets Its Own Interview | Data Puppets Bonus

On the Data Faces Podcast, I usually interview someone with a whole face. For this bonus segment, I made an exception. Scott Taylor's Data Puppets character "A-Eye" joins the show fresh from the Gartner Data & Analytics Summit. The puppet had thoughts about agentic AI, data governance, and vendors who couldn't spell AI two years ago. Key moments:1- A-Eye reports from the Gartner show floor2- "Agents writing code, reviewing code, deploying code, and then apologizing for the code"3- "AI is the Ozempic for data governance, baby"4- The Old MacDonald data anthem (yes, I sang along) Watch the full Episode 35 conversation with Scott Taylor:Blog: https://tinytechguides.com/blog/truth-before-meaning-the-three-word-fix-for-data-management/ [https://tinytechguides.com/blog/truth-before-meaning-the-three-word-fix-for-data-management/] Connect with Scott Taylor:LinkedIn: https://www.linkedin.com/in/scottdtaylor/ [https://www.linkedin.com/in/scottdtaylor/]MetaMeta Consulting: https://www.metametaconsulting.com/ [https://www.metametaconsulting.com/]Data Puppets: https://www.linkedin.com/company/data-puppets/ [https://www.linkedin.com/company/data-puppets/] Drop your thoughts in the comments!Like, share, and subscribe for more data and AI conversations. #DataPuppets #DataFacesPodcast #DataManagement #AI