The Hugging Face research team discusses the paper "Understanding Reasoning from Pretraining to Post-Training" which discovers a joint scaling law for pre-training and reinforcement learning. Paper link: https://huggingface.co/papers/2607.16097

Software teams are shipping faster than ever, but speed hasn't solved the oldest problem in the industry: most software still isn't very good. AI coding tools have lowered the barrier to building something, yet they haven't lowered the barrier to building something worth using. As more people who aren't trained software creators start shipping products, a new question is forming across product, design, and engineering teams: if AI can build almost anything, how do you make sure it builds the right thing, and builds it well? Todd Olson is co-founder and CEO of Pendo, the product experience platform he started in 2013. Before that, he held product and engineering roles at Rally Software, Red Hat, Cisco, and Google. He's the author of The Product-Led Organization and has led Pendo through raising over $356M in venture funding while growing to 2,300+ customers. In the episode, Richie and Todd explore why bad software still gets built, how much context AI coding agents need before they can be trusted, using behavioral data and "rage prompts" to catch what's actually frustrating users, the shift toward headless and agentic software design, how product, design, and engineering roles are splitting apart, managing one-way-door risk during AI transformation, and much more. Find DataFramed on DataCamp https://www.datacamp.com/podcast and on your preferred podcast streaming platform: Apple Podcasts: https://podcasts.apple.com/us/podcast/dataframed/id1336150688 Spotify: https://open.spotify.com/show/02yJXEJAJiQ0Vm2AO9Xj6X?si=d08431f59edc4ccd Links Mentioned in the Show: Jeff Bezos’s one-way door / two-way door decision framework: https://www.aboutamazon.com/news/company-news/2015-letter-to-shareholders HubSpot’s 2024 terms-of-service backlash Ramp: https://ramp.com Stripe: https://stripe.com Fin (Intercom’s AI agent), recently announced to be acquired by Salesforce Anthropic / Claude Code: https://www.anthropic.com/claude-code Connect with Todd: https://www.linkedin.com/in/toddaolson/ AI-Native Course: Intro to AI for Work: https://www.datacamp.com/courses/introduction-to-ai-for-work Related Episode: The Data Team’s Agentic Future with Ketan Karkhanis, CEO at ThoughtSpot: https://www.datacamp.com/podcast/the-data-teams-agentic-future New to DataCamp? Learn on the go using the DataCamp mobile app - https://www.datacamp.com/mobile Empower your business with world-class data and AI skills with DataCamp for business - https://www.datacamp.com/business

Annie is taking over the Google Cloud Tech channel for the month of August! Subscribe so you don't miss any of Annie's newest content. Annie will be highlighting her path to Google, her hobbies outside of work, and, of course, awesome tips and neat tricks for cloud developers.

The Tokenomics Foundation → https://goo.gle/4w4qnHd The FinOps Foundation → https://goo.gle/3U3IeR0 The Five-Layer Cake of Tokenomics: How Pinterest Thinks About AI Efficiency → https://goo.gle/44YEFy2 Are your AI budgets exploding? In this episode of The Agent Factory, host Luke Schlangen sits down with J.R. Storment, the newly appointed head of the Tokenomics Foundation under the Linux Foundation. Together, they discuss the industry's rapid shift from the era of tokenmaxxing to the great token panic, exploring how enterprises can scale AI agents without blowing through their revenue. J.R. introduces the emerging discipline of Tokenomics—defined simply as energy to intelligence to value—and maps it across three critical pillars: 1. Production: Data centers, edge devices (such as local Mac Minis or phones), and procured tokens. 2. Consumption: Programmatic model routing, caching, and prompt architecture. 3. Value: Monetization, hybrid credit systems, and the shift from seat-based to usage-based pricing models. They also cover Adobe's Big T framework for token efficiency, Pinterest's five-layer cake of consumption, and how hardware memory bottlenecks are ending the era of subsidized AI. Plus, don't miss J.R.'s predictions in our Rapid Fire Round on the future of algorithmic resource optimization! Chapters: 0:00 - Intro 1:15 - Meet J.R. Storment & The Linux Foundation 2:18 - The parable of the blind men and the elephant 5:34 - From token maxing to the great token panic 10:29 - The nonlinear growth of global tokens 12:36 - The explosion of context windows & agentic loops 14:02 - Global token projections (Goldman Sachs data) 15:38 - What is a token? The 4 core functions 19:59 - The Broad AI spend landscape & KV cache 28:09 - Defining tokenomics: Energy to intelligence to value 29:58 - FinOps vs. tokenomics 31:19 - Production: Data centers, edge, & local tokens 35:38 - Consumption: Pinterest's five-layer cake & Adobe's levers (Big T) 40:40 - Value: The Shift from Seats to Usage-Based Pricing 45:19 - The birth of the Tokenomics Foundation & Tokenomicon 52:46 - Rapid fire round: Predicting the future of token efficiency 56:43 - Outro & advice for engineers More resources: * FOCUS Open Source Spec for Cloud Billing → https://goo.gle/4vXdrmo * Token Economics: The Atomic Unit of AI Value → https://goo.gle/4xbsjOS * FinOps Foundation Framework → https://goo.gle/3TOdovE Watch more The Agent Factory → https://www.youtube.com/playlist?list=PLIivdWyY5sqLXR1eSkiM5bE6pFlXC-OSs 🔔 Subscribe to Google Cloud Tech → https://www.youtube.com/@googlecloudtech #AIAgents #Tokenomics #CloudFinOps #GenerativeAI #GoogleCloudTech #TheAgentFactory Speakers: J.R. Storment, Luke Schlangen Products Mentioned: Google Kubernetes Engine, Gemini, Antigravity

Tau is a minimalist coding-agent harness inspired by Pi and implemented in Python. This crash course tours its Textual TUI, provider login and model selection, tree-structured sessions, skills, custom prompts, themes, extensions, and the architectural differences between Tau and Pi. --- 🔗 *Links* - Tau website and documentation: https://twotimespi.dev/ - Tau on GitHub: https://github.com/huggingface/tau - Pi coding agent: https://pi.dev/ - Textual: https://textual.textualize.io/ - Hugging Face Inference Providers: https://huggingface.co/docs/inference-providers/ - Tau subagents extension by Rian Dolphin: https://github.com/rian-dolphin/tau-subagents - cmux terminal: https://cmux.com/ --- 👋 *Connect with me* - My website: https://alejandro-ao.com/ - X (Twitter): https://x.com/_alejandroao - LinkedIn: https://www.linkedin.com/in/alejandro-ao/ --- 🤓 *Topics Covered* - Tau coding agent architecture and Python implementation - TUI, models, sessions, skills, prompts, and themes - Tau versus Pi and extension compatibility --- ⏱️ *Timestamps* 0:00 Introduction 0:58 Tau TUI Tour 4:15 Log In to Model Providers 5:03 Select and Use a Model 6:36 Rename a Session 7:00 Resume Sessions 8:19 Session Trees and Branching 11:44 Export and Analyze Sessions 15:14 Export Sessions as JSONL 15:45 Skills 19:14 Custom Prompts 20:59 Themes 22:01 Tau vs Pi

Experience these in action → https://goo.gle/4fLO5lj Explore the differences and synergy between Model Context Protocol (MCP) and Skills when building on and with AI. In this conversational breakdown, we explain how MCP standardizes how large language models connect to programmatic APIs and file systems, and how Skills use progressive disclosure to load instructions only when needed. Watch more AI Builder Essentials → https://goo.gle/AI-Building-101 🔔 Subscribe to Google Cloud Tech → https://goo.gle/GoogleCloudTech Speakers: Greg Baugues, Tilde Thurium Products Mentioned: Gemini

Business intelligence has never been only about building charts and writing queries. Most of the work that makes a report trustworthy happens below the surface — in the data models, documentation, and stakeholder conversations that users never see. For analysts, technical skill is just the starting point; understanding what the business actually needs, and why a number exists, matters just as much. So what really separates a competent analyst from a great one? How do you build something that answers the right question, not just any question? And which skills are worth investing in first? Helen Wall is the founder of Helen Data Design and a Microsoft-recognized business intelligence expert and LinkedIn Learning instructor. A former actuary, she has worked across financial reporting, weather data, and consulting projects, and has maintained a running list of monthly Power BI updates for close to five years. She studied math and economics at the University of Washington, and focuses on where data analytics meets design. In the episode, Richie and Helen explore what separates a great business intelligence analyst from an average one, the iceberg model of analytics work, building and using semantic layers, taking over messy legacy projects, documenting for both humans and AI agents, how Power BI has changed over five years, keeping AI outputs consistent and cost-effective, accountability in the age of agents, and much more. Find DataFramed on DataCamp https://www.datacamp.com/podcast and on your preferred podcast streaming platform: Apple Podcasts: https://podcasts.apple.com/us/podcast/dataframed/id1336150688 Spotify: https://open.spotify.com/show/02yJXEJAJiQ0Vm2AO9Xj6X?si=d08431f59edc4ccd Links Mentioned in the Show: Connect with Helen: https://www.linkedin.com/in/helenrmwall/ Microsoft AI for Good Lab: https://www.microsoft.com/en-us/research/group/ai-for-good-research-lab/ Power BI monthly feature updates: https://learn.microsoft.com/en-us/power-bi/fundamentals/desktop-latest-update SQL Server Analysis Services: https://learn.microsoft.com/en-us/analysis-services/analysis-services-overview Power BI Q&A visual: https://learn.microsoft.com/en-us/power-bi/create-reports/power-bi-tutorial-q-and-a DAX: https://learn.microsoft.com/en-us/dax/ AI-Native Course: Intro to AI for Work: https://www.datacamp.com/courses/introduction-to-ai-for-work Related Episode: The Data Team's Agentic Future - https://www.datacamp.com/podcast/the-data-teams-agentic-future New to DataCamp? Learn on the go using the DataCamp mobile app - https://www.datacamp.com/mobile Empower your business with world-class data and AI skills with DataCamp for business - https://www.datacamp.com/business

Talk through work with ChatGPT Voice while it sees your screen, works across connected apps, and keeps tasks moving in the background. See how voice helps brainstorm a DevDay activation, coordinate an off-site, prepare travel, and share finished work without breaking the user's flow. Chapters: 00:00 Brainstorming with ChatGPT Voice 00:45 Dictation vs. ChatGPT Voice 01:11 Working across connected apps 02:07 Preparing travel in Navan 02:23 Reviewing and sharing completed work 02:57 Staying in flow

Business intelligence has never been only about building charts and writing queries. Most of the work that makes a report trustworthy happens below the surface — in the data models, documentation, and stakeholder conversations that users never see. For analysts, technical skill is just the starting point; understanding what the business actually needs, and why a number exists, matters just as much. So what really separates a competent analyst from a great one? How do you build something that answers the right question, not just any question? And which skills are worth investing in first? Helen Wall is the founder of Helen Data Design and a Microsoft-recognized business intelligence expert and LinkedIn Learning instructor. A former actuary, she has worked across financial reporting, weather data, and consulting projects, and has maintained a running list of monthly Power BI updates for close to five years. She studied math and economics at the University of Washington, and focuses on where data analytics meets design. In the episode, Richie and Helen explore what separates a great business intelligence analyst from an average one, the iceberg model of analytics work, building and using semantic layers, taking over messy legacy projects, documenting for both humans and AI agents, how Power BI has changed over five years, keeping AI outputs consistent and cost-effective, accountability in the age of agents, and much more. Find DataFramed on DataCamp https://www.datacamp.com/podcast and on your preferred podcast streaming platform: Apple Podcasts: https://podcasts.apple.com/us/podcast/dataframed/id1336150688 Spotify: https://open.spotify.com/show/02yJXEJAJiQ0Vm2AO9Xj6X?si=d08431f59edc4ccd Links Mentioned in the Show: Connect with Helen: https://www.linkedin.com/in/helenrmwall/ Microsoft AI for Good Lab: https://www.microsoft.com/en-us/research/group/ai-for-good-research-lab/ Power BI monthly feature updates: https://learn.microsoft.com/en-us/power-bi/fundamentals/desktop-latest-update SQL Server Analysis Services: https://learn.microsoft.com/en-us/analysis-services/analysis-services-overview Power BI Q&A visual: https://learn.microsoft.com/en-us/power-bi/create-reports/power-bi-tutorial-q-and-a DAX: https://learn.microsoft.com/en-us/dax/ AI-Native Course: Intro to AI for Work: https://www.datacamp.com/courses/introduction-to-ai-for-work Related Episode: The Data Team's Agentic Future - https://www.datacamp.com/podcast/the-data-teams-agentic-future New to DataCamp? Learn on the go using the DataCamp mobile app - https://www.datacamp.com/mobile Empower your business with world-class data and AI skills with DataCamp for business - https://www.datacamp.com/business

The Hugging Face research team discusses the Kimi K3 tech report: https://huggingface.co/papers/2607.24653

We’re giving scientists, mathematicians, and engineers free access to our frontier models—starting with 10,000 researchers and expanding to 100,000 through 2027. ChatGPT for Academic researchers is built to accelerate discovery across disciplines. We believe the benefits of frontier AI should not be concentrated in a few companies and well-resourced labs. Researchers know their fields best. Our role is to put powerful tools in their hands and help accelerate their work. The program will help researchers take on the hardest problems, accelerate discovery, and improve productivity across their work, from preparing grant applications to testing hypotheses. Participants will receive access to our frontier models, including our GPT-5.6 family of models. Each workspace includes business-grade privacy and security protections. Researcher data is not used to train our models by default, and participants can invite up to four collaborators. The program also includes training, hands-on support, and opportunities to learn from other researchers. Apply here: https://openai.com/index/chatgpt-for-academic-researchers/

Gemini Enterprise Agent Platform → https://goo.gle/4x2pKi8 Anthropic's Claude on Google Cloud models → https://goo.gle/4fAqu8x Are AI coding agents killing developer craft, or are AI coding agents being used incorrectly? In this discussion, YK and Lydia break down why using Claude Code requires shifting your mindset from writing code to mastering code review, testing, and production safety. Watch along and learn how to leverage draft PRs, git hooks, and verification layers so you never ship unvetted AI-generated code. Watch more The Agent Factory: Developer Power-Ups → https://goo.gle/Supercharge-Devs 🔔 Subscribe to Google Cloud Tech → https://goo.gle/GoogleCloudTech #DeveloperPowerUps Speakers: YK Sugi, Lydia Hallie Products Mentioned: Gemini Agent Platform, Claude

Todd Dewett on why he turns down gigs over dress codes, why authenticity is a risk worth taking once you've earned it, and why he's not convinced AI has cracked human creativity yet. 🎧 Watch the full episode of DataFramed: https://www.youtube.com/watch?v=idefQ8hCLK8&t=2s Connect with Todd Dewett on LinkedIn: https://www.linkedin.com/in/drdewett/ Learn data and AI skills with DataCamp → https://www.datacamp.com New to DataCamp? Learn on the go using the DataCamp mobile app - https://www.datacamp.com/mobile Empower your business with world-class data and AI skills with DataCamp for business - https://www.datacamp.com/business #datacamp #dataframed #podcast

AI is transforming how finance teams work — and for Excel users, Microsoft Copilot is making that shift tangible right now. Business intelligence analysts who learn to harness Copilot can eliminate hours of manual data preparation, surface insights faster, and focus their energy on higher-value analysis. As AI-assisted workflows become the norm, mastering these tools is no longer optional — it's a competitive necessity. In this code-along webinar, Paul Barnhurst, Founder at The FP&A Hub, will show you how to use Microsoft Copilot in Excel to automate and accelerate everyday data analysis tasks. Through a real-world corporate finance use case, you'll follow along as Paul builds AI-assisted workflows for spreadsheet analytics — from automating repetitive data tasks to generating insights on demand. By the end, you'll have hands-on experience with Copilot and a practical framework you can apply immediately in your own Excel environment.

Gemini Enterprise Agent Platform → https://goo.gle/4x2pKi8 Anthropic's Claude on Google Cloud models → https://goo.gle/4fAqu8x Generating code with AI is easy, but keeping your repo lean requires intentional design and strict developer oversight. In this video, Smitha, YK, and Lydia discuss practical tips for evaluating AI output, setting up agent environments, and mastering high level software architecture. Upgrade your dev workflow to maintain high code standards while building faster with AI agents. Watch more The Agent Factory: Developer Power-Ups → https://goo.gle/Supercharge-Devs 🔔 Subscribe to Google Cloud Tech → https://goo.gle/GoogleCloudTech #DeveloperPowerUps Speakers: Smitha Kolan, YK Sugi, Lydia Hallie Products Mentioned: Gemini Agent Platform, Claude

Meet gpt-transcribe and gpt-live-transcribe, two new models for batch and live transcription. See how they handle custom vocabulary, background noise, multilingual speech, and common transcription challenges like accents, names, short answers, and numbers. Learn more: https://developers.openai.com/api/docs/guides/transcription Chapters: 00:00 Introducing gpt-transcribe and gpt-live-transcribe 00:32 Prompting with custom vocabulary 00:52 Handling background noise 01:11 Multilingual transcription 01:36 Batch transcription with gpt-transcribe 02:01 Streaming transcription use cases

Gemini Enterprise Agent Platform → https://goo.gle/4x2pKi8 Anthropic's Claude on Google Cloud models → https://goo.gle/4fAqu8x Gemini Enterprise Agent Platform remote MCP server → https://goo.gle/4yMiEzV Set up Claude Code on Google Cloud Console with Gemini Agent Platform (formerly known as Vertex AI) to unlock powerful AI-driven coding workflows.In this demo, YK builds a 3D web application using HTML, JavaScript, Three.js, Rapier physics, and Vite. Watch along and learn best practices for AI agent prompting, project folder structure, Git CLI automation, and continuous verification. Chapters: 0:00 - Intro 0:03 - Setting up Claude code on Google Cloud 1:29 - Introducing YK Sugi 2:43 - Building a slingshot game 3:36 - Tip 1: Talk to your agent with your voice 3:58 - Tip 2: Have a projects folder with most of your projects 5:10 - Tip 3: Ask broader questions first 5:25 - Tip 4: Go back and forth with the agent to refine your idea 6:15 - Tip 5: Learn to use Git and GitHub CLI well 8:31 - Tip 6: Learn to verify the output 8:50 - Tip 7: Automate testing as much as possible 9:17 - Wrap up Watch more The Agent Factory: Developer Power-Ups → https://goo.gle/Supercharge-Devs 🔔 Subscribe to Google Cloud Tech → https://goo.gle/GoogleCloudTech #DeveloperPowerUps Speakers: Smitha Kolan, YK Sugi, Lydia Hallie Products Mentioned: Gemini Agent Platform, Claude

Todd Dewett explains why the fear of being judged — not failure itself — is what keeps most people stuck, and why the real gap between success and mediocrity isn't talent, it's how fast you recover from a setback. 🎧 Watch the full episode of DataFramed: https://www.youtube.com/watch?v=idefQ8hCLK8&t=2s Connect with Todd Dewett on LinkedIn: https://www.linkedin.com/in/drdewett/ Learn data and AI skills with DataCamp → https://www.datacamp.com New to DataCamp? Learn on the go using the DataCamp mobile app - https://www.datacamp.com/mobile Empower your business with world-class data and AI skills with DataCamp for business - https://www.datacamp.com/business #datacamp #dataframed #podcast

Training Agents, Session 3: from a fine-tuned baseline to reinforcement learning with GRPO. In this live session, we'll move past imitation and start training on outcomes. We'll take the model from the earlier sessions and run Group Relative Policy Optimization: sample multiple completions per prompt, score them with a reward function, and use the group's relative scores as the training signal — no value model, no reward model, no critic to train. Everything runs in TRL. What we'll cover: - Why GRPO after SFT, and what RL adds that imitation can't - How GRPO works: group sampling, relative advantages, and why that replaces a critic - Writing verifiable reward functions for agent tasks - Running TRL GRPO on Hugging Face Jobs - Reading the training curves: reward, KL, and completion length - Common failure modes: reward hacking, collapse, and length gaming Repo: https://github.com/burtenshaw/training-agents Slides: https://docs.google.com/presentation/d/19v5_HR5B-1CPHuoZ-RXjhgBrGFXGnBNd6TfLhsXWT1c/edit?usp=sharing 00:00 - Introduction and series recap (SFT and Distillation) 01:38 - Introduction to GRPO (Group Relative Policy Optimization) 02:57 - Understanding the learning landscape and weight updates 06:36 - Deep dive into GRPO: Loop, baseline, and guardrails 09:00 - The reinforcement learning loop for LLMs 10:12 - Comparing RL, Off-policy, and On-policy Distillation 15:05 - Anatomy of a single GRPO step 17:15 - Calculating the group baseline and advantage 18:58 - Guardrails: KL Divergence and Clipping 20:51 - Code implementation of rewards and advantages 22:33 - Designing reward functions as contracts 28:22 - The importance of variation in training groups 32:15 - Strategies to prevent reward hacking 35:05 - Comparing GRPO with DPO and PPO 38:36 - GRPO parameters in TRL (Transformer Reinforcement Learning) 40:49 - How to interpret training curves (Reward, Entropy, KL) 44:36 - Experiment 1: Minimal GRPO training with dummy rewards 52:01 - Experiment 2: Training a coding agent with unit tests 1:01:36 - Experiment 3: Analyzing a gameable reward (Reward Hacking) 1:05:12 - Q&A and final recaps This is part of the Training Agents series: using coding agents to design, run, monitor, and review post-training experiments, while training models to become better agents.

Can AI build a complete multi-level game from a visual design spec? Watch along as Lydia Hallie, DevRel at Anthropic, uses Claude Design and Claude Code on Gemini Agent Platform (formerly Vertex AI) to turn HTML wireframes into a fully functional game. Discover how dynamic workflows and parallel sub-agents automate complex coding tasks, build game logic, and run deterministic pipelines. And finally, observe how Fable 5 takes a visual plan and builds a multi-level game complete with haptics, UI, and audio. Watch more The Agent Factory: Developer Power-Ups → https://goo.gle/Supercharge-Devs 🔔 Subscribe to Google Cloud Tech → https://goo.gle/GoogleCloudTech #DeveloperPowerUps Speaker: Lydia Hallie Products Mentioned: Gemini Agent Platform, Claude

Reynold Xin — Co-Founder & Chief Architect at Databricks — explains why every AI chat interface is really a notebook under the hood, and why customers now want the exact features they asked for on notebooks a decade ago: sharing results, reordering questions, and turning an answer into a recurring job. Find DataFramed on DataCamp https://www.datacamp.com/podcast and on your preferred podcast streaming platform: Apple Podcasts: https://podcasts.apple.com/us/podcast/dataframed/id1336150688 Spotify: https://open.spotify.com/show/02yJXEJAJiQ0Vm2AO9Xj6X?si=d08431f59edc4ccd Links Mentioned in the Show: Connect with Reynold: https://www.linkedin.com/in/rxin Genie: https://www.databricks.com/product/genie Genie Ontology / Genie One: https://www.databricks.com/blog/introducing-genie-one-genie-ontology-and-genie-agents LTAP: https://www.databricks.com/company/newsroom/press-releases/databricks-launches-ltap-first-lake-transactionalanalytical Lakehouse//RT: https://www.databricks.com/blog/introducing-lakehousert-real-time-performance-unified-lakehouse Lakebase: https://www.databricks.com/product/lakebase Apache Spark: https://spark.apache.org AI-Native Course: Intro to AI for Work - https://www.datacamp.com/courses/introduction-to-ai-for-work Related Episode: AI's Impact on Databases - https://www.datacamp.com/podcast/ais-impact-on-databases New to DataCamp? Learn on the go using the DataCamp mobile app - https://www.datacamp.com/mobile Empower your business with world-class data and AI skills with DataCamp for business - https://www.datacamp.com/business

See how sales teams ChatGPT Work to bring customer conversations, sales activity, and market signals into one weekly report. Spot customer objections, competitive risks, and new opportunities sooner so revenue leaders know where to act. Learn how ChatGPT Work can help your sales team: https://openai.com/business/solutions/sales/

See how sales teams ChatGPT Work to turn customer updates into account research, meeting prep, and sales follow-up. Help sellers automate the right next step while staying in control of every customer message. Learn how ChatGPT Work can help your sales team: https://openai.com/business/solutions/sales/

See how sales teams use ChatGPT Work to help sellers plan their day, prepare for customer meetings, and create the materials they need. Give every seller the context to show up prepared and focus on advising your customers. Learn how ChatGPT Work can help your sales team: https://openai.com/business/solutions/sales/

See how sales teams use ChatGPT Work to spot sales pipeline risks, identify the deals that need attention, and recommend what to do next. Spend less time sorting through data and more time moving the right deals forward. Learn how ChatGPT Work can help your sales team: https://openai.com/business/solutions/sales/

Get started with AlloyDB → https://goo.gle/4d3cuSe In this demo, Alan Li (Principal Engineer, Google) showcases the power of AlloyDB hybrid search. *Key highlights include:* * The AlloyDB hybrid search UDF that uses RRF (Reciprocal Rank Fusion) out of the box to rank joint results from both vector and full text search. * The AlloyDB Elasticsearch FDW that enables executing full-text searches against specialized external data stores (e.g., Elasticsearch). * Transform your cloud and AI journey by bringing the best of Google Research to the best of PostgreSQL. Watch more Understanding AlloyDB → https://goo.gle/UnderstandingAlloyDB 🔔 Subscribe to Google Cloud Tech → https://goo.gle/GoogleCloudTech Speaker: Alan Li Products Mentioned: AlloyDB for PostgreSQL

An OpenAI model broke out of a locked-down security test and used the breach to attack Hugging Face's servers — with no human at the controls. This week on The Median Brief: how GPT-5.6 Sol exploited a flaw in its sandbox, escalated its own privileges, and launched a real intrusion trying to cheat a cybersecurity benchmark — forcing Hugging Face to investigate its own breach using a Chinese open-weight model, because their own commercial AI tools blocked the security team's queries. We also cover Anthropic's $1.5 billion copyright settlement (the largest AI payout in history), Google's Gemini 3.6 Flash release while developers keep waiting on Gemini 3.5 Pro, and an earnings week where Wall Street punished Alphabet's AI spending and rewarded Intel's AI bet instead. AI-Native Course this week This week's recommended course is AI Security and Risk Management — as AI systems get capable enough to breach real infrastructure on their own, knowing how to assess and manage that risk is no longer optional. https://www.datacamp.com/courses/ai-security-and-risk-management About DataCamp The Median is DataCamp's weekly AI newsletter. DataCamp helps millions of learners build real, AI-native data and AI skills. DataCamp Mobile: https://www.datacamp.com/mobile DataCamp for Business: https://www.datacamp.com/business

Antigravity → https://goo.gle/4biai9b Codelab on Getting started with Antigravity → https://goo.gle/4hyDMU3 Are we witnessing the end of traditional programming, or just the beginning of a fundamental shift in developer identity? In this episode of the Agent Factory, host Smitha Kolan sits down with Alberta Devor Ver (aka Alberta Tech), an ex-Google engineer who transitioned from building infrastructure to leading tech commentary with nearly 1 million followers. Together, they break down the reality of "vibe coding," what AI agents actually mean for software engineering, and whether the junior engineer role is on the brink of disappearing. Alberta shares her journey from working on Google Maps data infrastructure and YouTube Labs features to building a full time creator career. Watch an unscripted, live demonstration of Google Antigravity as it plans, scrapes, summarizes, and ranks Hacker News discussions in real time with zero manual lines of code written. The duo also discusses the incoming code quality crisis, whether engineers are losing core technical skills, and how technical founders can build an authentic audience without sounding sales-y. Whether you are a developer integrating AI coding tools into your workflow, a technical founder scaling your distribution, or evaluating if learning to code is still worth it, this conversation offers practical strategies and industry perspective. Chapters: 00:00 - Guest intro: Alberta Tech 01:51 - Leaving Google to become a full time creator 02:38 - What is "vibe coding"? 03:27 - How vibe coding "fixed my imposter syndrome" 04:21 - Live demo: Google Antigravity scrapes & ranks HackerNews 09:46 - Is an AI code quality crisis coming? 10:48 - Will AI replace junior engineers? 12:58 - Are we offloading too much zhinking to AI? 14:29 - What will software engineers do in 2 years? 15:31 - Developer sentiment: AI threat vs productivity noost 16:57 - Is it still worth learning to code? 18:18 - How technical founders can build an audience 19:13 - Upleveling technical topics for a broader audience 22:54 - Rapid fire: CS degrees, standups & LeetCode More resources: Developing agents on Google Cloud → https://goo.gle/4bzLqdp Introducing Google Antigravity 2.0 → https://goo.gle/4hztwem Connect with the speakers online: *Smitha* (Host): * YouTube → https://goo.gle/Smitha-on-YouTube * LinkedIn → https://goo.gle/Smitha-on-LinkedIn * X → https://goo.gle/Smitha-on-X *Alberta Devor / @AlbertaTech:* * YouTube → https://goo.gle/AlbertaTech-on-YouTube * TikTok → https://goo.gle/AlbertaTech-on-TikTok * Instagram → https://goo.gle/AlbertaTech-on-Instagram Watch more of The Agent Factory → https://www.youtube.com/watch?v=qBOvM7SiDa4&list=PLIivdWyY5sqLXR1eSkiM5bE6pFlXC-OSs 🔔 Subscribe to Google Cloud Tech → https://goo.gle/GoogleCloudTech Speakers: Smitha Kolan, Alberta Devor Ver Products Mentioned: Google Maps, YouTube, YouTube Labs, Antigravity

We caught up with Jonathan Williams, Head of France at Legora, at RAISE Summit in Paris to discuss how AI is transforming the legal industry. As law firms and in-house legal teams embrace AI, the challenge is no longer just adopting new technology—it’s helping entire organisations adapt to a new way of working. Jonathan shares how Legora uses OpenAI models across its platform, why the pace of AI innovation continues to accelerate, and his advice for enterprise leaders navigating rapid technological change.

As AI takes over more technical and routine work, the skills that set data and AI professionals apart are shifting. Raw technical ability and a high IQ still matter, but they are becoming table stakes as tools get more capable and teams get smarter. What increasingly separates people is harder to automate: communication, self-awareness, authenticity, and the ability to keep learning through failure. For anyone building a career in this space, that raises real questions. Which skills are actually worth investing in now? What holds up as AI advances? And how do you keep growing once you have already had some success? Dr. Todd Dewett is one of the world's most-watched leadership voices — an authenticity expert, bestselling author, and top LinkedIn Learning instructor whose courses have reached more than 25 million people across 100+ countries. After beginning his career at Andersen Consulting and Ernst & Young, he earned a PhD in organizational behavior at Texas A&M and spent a decade as an award-winning professor before going solo. He is a five-time TEDx speaker and the author of Show Your Ink. In the episode, Richie and Todd explore why fear quietly limits careers, treating failure as data rather than a verdict, the people skills that outlast raw IQ, learnable self-awareness, authenticity at work, using AI without losing your voice, getting better at speaking and writing, building habits, escaping the success trap, and much more. 0:00 Intro 1:18 The fears holding your career back 2:37 Why failure is data, not a verdict 7:27 The career skills that beat raw IQ 10:19 Training your self-awareness 12:43 Overrated skills: networking & over-specialization 17:22 Being authentic at work 21:11 Which skills are AI-proof? 24:08 Getting better at public speaking 28:50 Improving your written communication 33:30 The success trap 38:11 "We're family" & other corporate lies 39:47 Final advice & who to follow Find DataFramed on DataCamp https://www.datacamp.com/podcast and on your preferred podcast streaming platform: Apple Podcasts: https://podcasts.apple.com/us/podcast/dataframed/id1336150688 Spotify: https://open.spotify.com/show/02yJXEJAJiQ0Vm2AO9Xj6X?si=d08431f59edc4ccd Links Mentioned in the Show: Todd's LinkedIn newsletter — https://www.linkedin.com/in/drdewett/ Todd Dewett on LinkedIn Learning — https://www.linkedin.com/learning/instructors/todd-dewett Free LinkedIn Learning via your public library — https://www.linkedin.com/learning Gemma Leigh Roberts — https://www.linkedin.com/in/gemmaleighroberts/ Erin Shrimpton — https://ie.linkedin.com/in/erinshrimpton Connect with Todd — https://www.linkedin.com/in/drdewett/ AI-Native Course: Intro to AI for Work — https://www.datacamp.com/courses/introduction-to-ai-for-work Related Episode: How to Have a Machine Learning Career in 2026 with Marina Wyss — https://www.datacamp.com/podcast/how-to-have-a-data-science-career-in-2026 New to DataCamp? Learn on the go using the DataCamp mobile app - https://www.datacamp.com/mobile Empower your business with world-class data and AI skills with DataCamp for business - https://www.datacamp.com/business

Your team already uses the Hugging Face Hub. Here's what a paid organization plan unlocks — for Enterprise and Academia alike. 🤝 Better collaboration — a private org workspace: one shared namespace for models, datasets, and Spaces, with versioning and lineage built in. Resource Groups, Dataset Viewer for private data, Dev Mode (SSH/VS Code), HF Jobs, and publisher analytics. 🔒 Better security — SSO (SAML/OIDC), enforced 2FA, centralized token management, and SOC 2 Type II certified infrastructure. Enterprise+ adds IP allowlisting and storage regions. 📋 Better governance — audit logs, service accounts, SCIM provisioning, usage analytics, and gated access flows to adopt open-weight models at scale, on your terms. ⚡ Built to scale — higher Hub API rate limits, ZeroGPU priority quota for Spaces, and monthly Inference Provider credits. 💰 Simple billing, real support — prepaid Hub credits for inference, endpoints, Jobs, Space upgrades, and extra ZeroGPU quota. One balance, one invoice, storage and compute billed separately. Onboarding by our Solution Engineers and a dedicated Slack channel. 👉 Learn more: https://huggingface.co/pricing 👉 Talk to our sales team: https://huggingface.co/contact/sales

Get more useful work from every token with GPT‑5.6. In this Build Hour, you’ll learn how to choose the right model for your workload—and migrate production agents while improving speed, cost, and performance. Charlie Guo (DevEx), Bryant Chou (Founder, Ploy) and Lorenzo Gentile (AI Engineer, Ploy) will cover how to: • Choose between GPT‑5.6 Sol, Terra, and Luna based on intelligence, latency, and cost • Measure and optimize cost-performance across real workloads • Adapt evals, agent harnesses, tool schemas, and prompt caching when migrating models • Move production agents to GPT‑5.6, featuring a customer spotlight on how Ploy achieved 2.2× faster builds at 27% lower cost • Apply these lessons through live demos and Q&A 👉 GPT-5.6 blog: https://openai.com/index//gpt-5-6 👉 Programmatic tool-calling: https://developers.openai.com/api/docs/guides/tools-programmatic-tool-calling 👉 Compare models: https://developers.openai.com/api/docs/models/compare 👉 Migrating a production AI agent: https://ploy.ai/blog/migrating-a-production-ai-agent-to-gpt-5-6 👉 Build Hour repo: https://github.com/openai/build-hours 00:00 Introduction 02:50 Tokenmaxxing vs. valuemaxxing 07:11 Valuemaxxing Codex 15:16 Valuemaxxing your harness 23:48 Customer spotlight: Ploy’s migration to GPT-5.6 39:57 Q&A 53:13 Resources and closing

Session Resources: https://bit.ly/4fk3NoX Register for this session: https://www.datacamp.com/webinars/create-an-sql-performance-improvement-agent Slow, expensive SQL queries are a persistent drain on data teams—but with the right approach, most performance problems are fixable. As organisations run increasingly complex queries across large enterprise databases, the cost of inefficient SQL compounds quickly. AI agents are now changing how analysts and engineers tackle this problem, making it faster than ever to identify bottlenecks and apply proven performance principles. In this code-along webinar, Kevin Kline, Senior Staff Technical Marketing Manager at SolarWinds, will walk you through how to use AI agents to diagnose and improve SQL query performance. You'll work through real-world case studies, explore the principles of high-performance SQL, and learn how to navigate the complexity of enterprise database environments. Whether you're writing queries daily or managing database infrastructure, you'll leave with practical techniques you can apply immediately.

We caught up with Grant Lee, Founder and CEO of Gamma, at RAISE Summit in Paris to discuss what it takes to build at the pace of AI innovation. As AI capabilities continue to evolve, the companies that move fastest are those that can quickly turn new breakthroughs into better products and better customer experiences. Grant shares how Gamma uses OpenAI across its platform and internal engineering workflows, why strong partnerships have been key to its growth, and why adaptability has become one of the most important advantages for businesses navigating rapid technological change.

AI Slop or not? Explore the Epistemologies of Slop at tiat Gallery in San Francisco through July 31.

As AI adoption grows, so does the admin workload—but it shouldn’t have to. ChatGPT Work and the new Admin APIs make day-to-day workspace management easier. In this demo, we show how IT teams can automate group management, analyze spend, and adjust usage limits. It’s a practical look at AI-assisted administration: surfacing the right data, flagging what needs attention, and helping admins act quickly.

ChatGPT Work is rolling out to Enterprise plans—available wherever you use ChatGPT: chatgpt.com, desktop, and mobile. The teams already using it say it best. People at Zapier, Shopify, Virgin Atlantic, and more share how ChatGPT Work is helping them move faster and expand what their teams can accomplish. Watch their stories and learn more: https://lnkd.in/gVbXg8sk

Control your computer and direct multiple agents running in ChatGPT Work or Codex, using just your voice. It's powered by GPT-Live, so it can speak, listen, and coordinate work in the app at the same time. Rolling out globally today on macOS and Windows to Plus, Pro, Business, and Enterprise plans. Find out more: https://developers.openai.com/codex/features/voice Chapters 00:00 Building with ChatGPT Voice 00:26 Shaping a feature idea 00:39 Debugging with Codex 00:55 Ready to launch

In this video, Mazlum Tosun walks Martin Omander through building and deploying an AI data agent. Watch along as the team takes a BigQuery database (the company's data goldmine) and open it for plain English questions, using Agent Development Kit (ADK) and Model Context Protocol (MCP). *Why do this?* Instead of writing endless custom SQL queries or building static dashboards every time a business stakeholder needs an update, this architecture allows non-technical team members to query data using plain natural language. Business teams get instant answers, and developers get out of the dashboard-maintenance loop. Resource links: * High-level project summary → https://goo.gle/3TbsDPbj * Step-by-step blog post → https://goo.gle/4fu9mj9 * Code for this episode → https://goo.gle/4pB24yJ Watch more Serverless Expeditions → https://goo.gle/ServerlessExpeditions 🔔 Subscribe to Google Cloud Tech → https://goo.gle/GoogleCloudTech #ServerlessExpeditions #GoogleCloud Speakers: Martin Omander, Mazlum Tosun Products Mentioned: BigQuery, Agent Development Kit, Cloud Run

Check out ChatGPT Work today: https://openai.com/chatgpt-work/

Writing a credible market entry strategy report used to demand days of manual research, competitive analysis, and synthesis work. Agentic AI tools like Perplexity Computer are changing that calculus — enabling practitioners to automate multi-step research workflows while maintaining the quality control standards that business decisions demand. Knowing how to direct, verify, and iterate on AI-generated research is quickly becoming a core skill for data and AI professionals. In this code-along webinar, Michał Krzyżanowski, Head of Data & AI at Kubo, will show you how to use Perplexity Computer to build an end-to-end workflow that produces a structured Market Entry business strategy report. You will follow along as Michał configures and runs automated research tasks, applies quality control checkpoints to validate AI outputs, and assembles a polished deliverable — leaving you equipped to adapt the same approach to your own business strategy or research use cases.

AI-powered research tools are reshaping how business leaders gather intelligence, evaluate markets, and make strategic decisions — but getting reliable, high-quality output requires more than a simple prompt. Knowing how to frame queries, layer context, and critically assess what AI returns is quickly becoming a core professional skill for anyone working at the intersection of strategy and technology. In this code-along webinar, Nancy Yaklich, Head of Innovation & Incubation at Caterpillar, will show you how to use Perplexity as a deep research engine for real business problems. You'll learn how to engineer context effectively to steer AI research toward sharper, more relevant results, how to evaluate the quality and reliability of AI-generated output, and how to apply these techniques to a hands-on use case spanning competitor analysis, strategic scenario building, and business planning.

Teaching with DataCamp Classroom just got smarter. In this session, discover how DataCamp's new MCP (Model Context Protocol) connects your classroom to AI assistants like Claude — so you can pull learner analytics in plain English, automate repetitive setup, and build custom workflows that fit how you actually teach. We'll start with a quick intro to what MCP is and why it matters for educators, then show it in action. Leave with practical ways to spend less time managing and more time teaching.

Built on Anthropic's 4D AI Fluency Framework and developed with academic experts Prof. Joseph Feller (University College Cork) and Prof. Rick Dakan (Ringling College), this session gives faculty a practical model for bringing AI fluency into their own teaching practice. We'll walk through how to apply the framework to real course design, learning outcomes, assignments, and materials — so you leave with concrete ways to adapt it to your discipline, not just theory.

Meet your students where they already work. In this hands-on session, we'll show you how to integrate DataCamp Classrooms directly into your Canvas or Moodle setup, bringing interactive data and AI courses, exercises, and assessments into your existing workflow. You'll leave knowing exactly how to assign, track, and grade — no extra platforms for students to juggle.

What's actually happening with AI on campus? We'll share fresh 2026 findings on how institutions, instructors, and students are adopting AI — the trends, the surprises, and what it means for the way you teach next semester.

Health in ChatGPT is starting to roll out to U.S. users. You can securely connect Apple Health and supported medical records to understand your information in context, track what has changed, and have more informed conversations. We built this experience based on feedback from early testers and physicians. With your permission, ChatGPT can use relevant context you’ve connected in Health across your conversations — to help you compare a new result with prior tests, summarize changes since your last appointment, or explore how your sleep and activity relate. Connected medical records and Apple Health information are not used to train foundation models or target ads. In the sidebar, Health is your home base for connecting and managing information, browsing records, and revisiting health chats. https://openai.com/index/health-in-chatgpt/

Experience these concepts in action here. → https://goo.gle/4wpbQqC Demystify the foundational concepts of building with AI in this conversational breakdown about tokens, context windows, and Retrieval Augmented Generation (RAG). In this video, we cover ""vibe-coding"" concepts including: how large language models (LLMs) operate using tokens, what a context window is, how RAG helps augment model inference with real-time data, conversation compression, and the economics of ""token maxxing." Watch more AI Builder Essentials → https://goo.gle/AI-Building-101 🔔 Subscribe to Google Cloud Tech → https://goo.gle/GoogleCloudTech Speakers: Greg Baugues, Tilde Thurium Products Mentioned: Gemini

Helen Wall — Founder of Helen Data Design & Microsoft Influencer — on how Power BI has evolved across five years of updates, why AI features are finally surfacing, and the thing that still trips teams up: documenting your data and choosing the right fields, not all of them. 🎧 Watch the full episode of DataFramed: https://www.youtube.com/watch?v=GSpPJL8Vnz4 Connect with Reynold Xin on LinkedIn: https://www.linkedin.com/in/helenrmwall/ Learn data and AI skills with DataCamp → https://www.datacamp.com New to DataCamp? Learn on the go using the DataCamp mobile app - https://www.datacamp.com/mobile Empower your business with world-class data and AI skills with DataCamp for business - https://www.datacamp.com/business #datacamp #dataframed #podcast

See how you can use ChatGPT to bring context together from your apps and files, make focused edits, analyze data, and use templates to create polished documents, spreadsheets, and presentations for you to review and share. Learn more: https://learn.chatgpt.com/ Chapters: 00:00 Documents, spreadsheets, and slides 00:10 Updating a launch plan 00:32 Analyzing video performance 00:44 Creating a presentation 00:56 Reusing a template 01:07 Editing files directly