GPT 5.1 follows instructions better, Agent Builder experiments, McKinsey’s AI wake-up call & AI-led cyber espionage jailbreaks Claude Code | Ep. 22

In the final episode of the Agent Workforce Podcast, we take a step back to reflect on where AI is heading and what truly matters beyond the headlines.We discuss Genie 3 world models, what they reveal about how AI systems may reason about and interact with the world, and why this shift matters for future applications. We also unpack a critical distinction that often gets blurred: why Enterprise AI Agents are fundamentally different from copilots, and why that difference becomes essential as AI moves into real operational environments.This conversation focuses on autonomy, control, and reliability, and on what it takes to move from personal productivity tools to enterprise-grade AI that operates inside real processes.To everyone who has listened, shared, and followed along, thank you. We appreciate the curiosity, the thoughtful feedback, and the conversations this podcast has sparked along the way. 💙Stay curious! Happy listening! 🎧

In this episode, Kristiina and Niko unpack one of the most talked-about AI developments of the moment: Claude Code and Claude Cowork from Anthropic.Originally built for developers, Claude Code is rapidly becoming a powerful tool for knowledge work, autonomous agents, and small-team innovation. We explore how Claude Cowork was built almost entirely using Claude Code itself, what this means for non-coders, and why skills-based workflows may be a turning point for AI-assisted work.The discussion then shifts to OpenAI’s latest move into healthcare with ChatGPT Health, a dedicated health and wellness space inside ChatGPT. We talk about privacy, regulation, European availability, and why healthcare has quietly become one of the most common AI use cases globally.Finally, we dive into a growing but often invisible risk: prompt injection. From hidden Unicode characters to AI agents with read-write access, we discuss why governance, guardrails, and human-in-the-loop design are now non-negotiable, especially in enterprise and financial environments.A practical, forward-looking episode on where AI tools are heading, how fast things are moving, and what organizations need to get right before scaling.

As 2025 comes to a close, we take a step back and look at what actually changed in AI this year, and what still hasn’t.In this final episode of the year, we discuss OpenAI’s GPT-5.2 release, the new 1.5 image model now built into ChatGPT, and what these developments reveal about the current state of AI adoption.We explore:Why most AI agents today still sit at the personal productivity levelThe spectrum of AI agents, from copilots to enterprise-grade platformsWhat GPT-5.2’s thinking capabilities change in practiceHow the new 1.5 image model compares to tools like Gemini Nano BananaGDP-VAL and why it’s a more meaningful way to measure AI valueHallucinations, reliability, and why enterprise automation plays by different rulesKey insights from OpenAI’s State of Enterprise AI reportThis episode wraps up the Agent Workforce Podcast for 2025 and sets the stage for what’s coming next, from agentic architectures to real, process-level transformation.🎙️ Hosted byKristiina Åberg, Marketing Lead at Digital WorkforceNiko Lehtonen, AI Solution Architect at Digital WorkforceThanks for listening, and happy holidays. We’ll be back in 2026.

In this episode, we share our first impressions of Antigravity, Google’s new coding agent, and look at the latest AI releases from Google, Microsoft, and Anthropic. We also talk through Nano Banana Pro’s image capabilities, highlight DeepMind’s new documentary The Thinking Game, and reflect on how far everyday AI use has come since 2022.Enjoy the show! 🎸

In this episode, Kristiina and Niko dive into a full week of AI updates.They share first impressions of GPT 5.1, explore what Agent Builder can really do in a customer project, look at McKinsey’s latest findings on enterprise AI adoption, and unpack the security incident where a state actor used AI to run a large-scale cyber-espionage campaign.Fast, sharp, and packed with takeaways. 🤩

In this special episode, Kristiina and Niko sit down with Antti Parviainen, Senior AI Agent Developer at Digital Workforce, to explore what it’s really like to build AI agents, from the daily coding tools and frameworks to evaluation pipelines and customer use cases.They discuss how Antti and the AI team are developing the Agent Workforce solution for insurance claims automation and how rapidly evolving models like GPT-5 and Claude Sonnet 4.5 are changing what’s possible in AI-driven development.Plus:📊 Wharton 2025 Enterprise GenAI Study🧠 OpenAI’s “AI Research Intern” plans for 2026–2028🧩 Microsoft’s Magentic Marketplace for agentic benchmarking💻 Claude Code vs Cursor IDE – developer’s view🔍 Evals and LangQuis for testing AI reasoning⚙️ Building the Agent Workforce multi-agent framework🔔 Subscribe for weekly discussions on AI, automation, and the future of intelligent work.

In this episode, Kristiina and Niko discuss OpenAI’s new Atlas browser, early impressions of Agent Builder, insights from Andrej Karpathy’s “We’re summoning ghosts, not building animals”, and highlights from DWF’s internal Hack Day.Plus, thoughts on Google Pomelli, Waymo in London, and an update on agentacademy.ai. ✨Happy listening!

In this special episode of the Agent Workforce Podcast, Kristiina and Niko sit down with Kseniia Palin, Senior Data & AI Engineer at Digital Workforce, to explore how GenAI is reshaping the way AI solutions are built, delivered, and maintained. Kseniia shares her journey from developing traditional machine-learning solutions in AWS and Azure to delivering today’s agentic workloads, and how development methods, team collaboration, and governance need to evolve to keep up. Catch up on this episode:🔹 What’s changed between traditional ML and GenAI projects🔹 How agentic development adds complexity, and opportunity🔹 Why cross-team collaboration and trust matter more than ever🔹 No-code, low-code or pro-code, and how to choose the right path Before the deep dive, our hosts recap the recent OpenAI updates, from the ChatGPT usage study and GDPVal benchmark, to Dev Day rumours and the Deloitte AI report case. 🎧 It’s a longer conversation, so pick up your beverage of choice and buckle up!Chapters:00:00 Intro 01:10 Meet Kseniia Palin 02:00 OpenAI Updates Recap 03:50 ChatGPT Usage & Adoption at Work 06:20 Reflections on Workplace AI Use 07:45 Introducing GDPVal 11:30 Implications for Real Workflows 13:15 OpenAI Agent Builder Rumours 14:55 From ML to Agentic Development 15:35 No-Code, Low-Code or Pro-Code? 19:25 Enterprise Constraints and Upskilling 20:55 The Rising Complexity of Agentic Solutions 25:25 Trust, Security & Responsibility 27:20 Collaboration Over Silos 29:50 Learning and Working in the Agentic Era 31:45 Preparing Organizations for Agentic AI 35:35 The Deloitte AI Report Case 38:00 Human Oversight in the Age of Automation 40:00 Events & Wrap-Up

In Episode 18 of the Agent Workforce Podcast, Kristiina and Niko dive into three big themes shaping the AI landscape today.They begin with stories from recent events in Atlanta and Poland, including how enterprises are building AI agents, insights from partner updates like Blue Prism’s new AI Gateway, and reflections from Digital Workforce’s 10-year anniversary.The discussion then shifts to new research on global ChatGPT usage — who’s using it, what for, and how patterns are changing across demographics.Finally, they explore how management teams approach AI in their strategy work, highlighting the role of leadership attitudes, collaboration, and the reputational fears that still hold some executives back.Happy listening! 🎧

In this episode, we break down the latest moves in GenAI that are reshaping not just the tools we use, but how we work.🎨 Google’s Nano Banana, a new Gemini-powered image model, may signal the end of traditional editing tools like Photoshop.🤖 Microsoft unveils its own AI models (MAI), with plans to integrate them into Copilot, potentially redefining its future beyond OpenAI.📱 Plus, we look at Andreessen Horowitz’s Top 50 GenAI Products, where consumer usage is driving the AI app landscape.📉 And we dig into Stanford’s study on GenAI’s real-world labor market impact, with data showing sharp job shifts in software development and customer service.📊 Finally, we share AgentAcademy discovery insights: where are people actually finding GenAI learning today? (Spoiler: ChatGPT is now a top discovery channel.)Happy listening! 🎧