The Agentic AI Gap: When Tech is Used Before its Architecture is Ready - Anthony Alcaraz, Agentic AI Architect (EP 25)

Tech and Democracy: How Can Both Be Connected to Create Trust? with Nexus Politics (EP 27)

🎙️ with Magnus Strobel, Co-Founder and CEO of Nexus Politics Trust in politics has been eroding across Western democracies for over a decade, and Magnus Strobel thinks the failure is in how democracy works, in the process that has stopped feeling participatory. His company, Nexus Politics, is a for-profit platform built to map the distance between what citizens actually think and what politicians actually do - and to make that distance impossible to ignore. 🔍 Episode overview This is a conversation about whether transparency can rebuild participation once the machinery of democracy has stopped feeling participatory. It is also about a quieter problem: how a founder building a trust instrument decides whether anyone actually trusts it. Magnus Strobel and his team create an architecture for a digital democracy platform: how citizen opinion gets routed to the right political actors, how the system maps public sentiment in real time, and where accountability is supposed to live. The harder questions arrive underneath: Why build this as for-profit rather than not-for-profit, and why that choice is the one that makes political neutrality credible. What politicians say they want from such a tool, and why their enthusiasm might mean less compared to how they use it specifically. It is a founder's conversation that keeps circling back to a single uncertainty: you can build the mechanism for trust, but you cannot yet prove the trust is there. ⚖️ Key themes * Why the crisis is in how democracy functions, not in democracy itself - and what that distinction changes * How a for-profit structure becomes the argument for political neutrality * Mapping the gap between what voters think and what politicians do * What politicians actually want from civic tech, and why positive feedback is the hardest signal to trust * Tech as a tool that can repair democratic trust or deepen the damage, depending on who uses it and how 🤝 About the guest Magnus Strobel is co-founder of Nexus Politics, a digital democracy platform built to rebuild participation and accountability in representative democracies. His background is in behavioral economics, which surfaces throughout the conversation in his attention to the gap between what a system is designed to do and what people actually do with it. He builds from Munich, embedded in the local startup ecosystem, with a stated ambition modelled partly on Taiwan's experience of using participation tools to lift satisfaction with democracy. 🌍 Chapter markers * [00:09] What comes to mind when a democracy founder thinks about trust * [02:59] Opening the fragmented machinery of politics - participation, transparency, accountability * [05:59] Why for-profit is the route to credible neutrality * [16:08] The hardest part is always reality - and what politicians really want * [22:49] Can tech rebuild democratic trust, or does it cut both ways * [35:48] In-between moments: trust, division, and where a founder sits right now ⛓️‍💥 Links * Nexus Politics:  www.nexuspolitics.org [http://www.nexuspolitics.org/] * Magnus Strobel LinkedIn: https://www.linkedin.com/in/strobelmagnus/ * Audrey Tang / Taiwan digital democracy: https://www.demnext.org/people/audrey-tang * Rebuild conference, Copenhagen: https://www.rebuild.net * Related episode - Rebuilding Trust: Tech, Politics and Entrepreneurial Leadership (EP 06)

AI in China and in Europe: Trust, Differences, and Future Implications - Vincent Xiang, Founder China AI Connect (EP 26)

Europe and China are on different AI paths at different speeds. Vincent Xiang has spent years inside that corridor: He has been working as a translator between Chinese AI founders and European investors and corporates, and this conversation dives into his experiences, conversations, and operations on the ground and in-between. 🧭 Episode overview European executives are excited about Chinese AI momentum. But they're also stuck before they act. Chinese founders interpret some of Europe's regulations as inefficiency. Both sides are operating with simplified labels that are accurate enough to feel right and wrong enough to produce bad decisions. Vincent walks through what he actually sees on the ground - why trust in China gets delegated to systems rather than built between strangers, why "AI superpower" and "surveillance dystopia" both miss the territory, why fragmentation is now treated as permanent reality by founders, and what European companies serious about engaging China should do before they book a single meeting. 🔍 Key themes discussed * The different first questions Europe and China ask about new technology, and what each one produces downstream * Trust as delegated infrastructure - the Alipay escrow story and why people trust the system rather than the strangers in it * Why both Western labels for Chinese AI are wrong in the same direction, and what gets missed when leaders operate with them * The three-layer coordination of government, platforms, and institutions in China, and what its absence looks like in Europe * Fragmentation as the new permanent reality, and why compliance has to be built in as a product feature from day one 👤 About the guest Vincent Xiang is the founder of China AI Connect, a research and advisory practice helping European investors and corporates evaluate whether Chinese AI is relevant to their strategy, and helping Chinese founders understand the European market. He lived in Germany for seven years, writes the China AI Connect briefings on Chinese AI and deep-tech policy and players, and organises executive trips that bring European leaders to meet founders and operators on the ground. His vantage point is one of the few that sits genuinely between the two systems. ⏱️ Chapter markers [00:55] The first word that comes to mind: difference [05:00] People trust the system, not the strangers in it [12:01] Why "AI superpower" and "surveillance dystopia" both miss the territory [19:00] Three layers of coordination: government, platforms, institutions [22:30] Fragmentation as permanent reality, and compliance as a product feature [35:00] The robotics inflection and what favourable policy makes possible 🔗 Links Vincent Xiang on LinkedIn - https://www.linkedin.com/in/yxiangeclille/ China AI Connect on Substack - https://vincentxiang.substack.com AI 2030 / AI Plus initiative reference - https://www.fmprc.gov.cn/eng/xw/zyjh/202509/t20250924_11715960.html Related episode - Episode on Trust as Geopolitical Requirement: Eva's WEF 2026 recap - https://open.spotify.com/episode/1RKtxdJWXcQH8vnpnDtgEP?si=u_MfnmOvQ2-AXSPRONX6Gw

The Agentic AI Gap: When Tech is Used Before its Architecture is Ready - Anthony Alcaraz, Agentic AI Architect (EP 25)

Most enterprises have the technology to run agentic AI. They do not yet have the data architecture, identity layer, or empowered workforce to actually trust it. Anthony Alcaraz argues that the bottleneck for agentic AI has shifted from building the agents to building everything around them — and that the organisations most at risk are the ones keeping a human in the loop and calling it transformation. This conversation is for leaders sitting between AI pilots that worked and production systems that have not yet arrived. 💡Episode overview Anthony joins Eva to map what changes when AI shifts from reactive systems to agents that observe, reason, and act. The conversation moves through what enterprises miss in their own data — systems of record that capture what happened but not why — and the new attack surfaces agents introduce, including tool poisoning. Anthony names the empowerment gap inside organisations: business experts who hold the knowledge agents need, with no clear path to building anything themselves. The most provocative moment lands near the end, when Anthony argues that human-in-the-loop adoption can be a way of avoiding actual transformation rather than achieving it. 🔍 Key themes discussed * The shift from reactive to agentic systems, and what trust has to carry now * Why most enterprise data is missing the why behind decisions * Tool poisoning and the new attack surface for agents * The empowerment gap between business knowledge and technical capability * Graph architecture as the control layer for agentic reasoning * Why human-in-the-loop can be a refusal to transform 👤 About the guest Anthony Alcaraz works across three vantage points that rarely sit together: he architects agentic AI systems, invests in early-stage AI startups as an angel, and is the author of Agentic Graph RAG with O'Reilly. He spends most weeks in conversation with founders attempting to enter regulated enterprises, and most evenings building software with the same tools he writes about. His perspective on this episode comes from watching the same gap repeat itself across organisations of very different sizes — the technology is ready, and most of the systems around it are not. 📍 Chapter markers * [00:00] What changes when AI moves from reactive to agentic * [05:42] Why agents need access — and what enterprises have not built * [10:29] The three problems: data, governance, and the people in between * [23:13] Graph architecture and the missing why of enterprise data * [32:06] The empowerment gap that no one has solved yet * [45:17] In-between: where Anthony finds himself now 🔗 Links * Anthony Alcaraz LinkedIn — https://www.linkedin.com/in/anthony-alcaraz-b80763155/ * Agentic Graph RAG (O'Reilly) — https://www.oreilly.com/library/view/agentic-graphrag/9798341623163/ * Foundation Capital context graph thesis — https://foundationcapital.com/ideas/the-case-for-context-graphs * Related episode — Trust as an operating system in AI companions https://open.spotify.com/episode/5t4BtgevPOtMWUfB4jThWX?si=oGo2JPHNTeCTxbqkNXDJMw * Eva Simone Lihotzky's LinkedIn: https://www.linkedin.com/in/evalihotzky/

Why AI Makes Political Authenticity Harder to Trust – Dr. Michael Cohen (EP 24)

AI has collapsed the cost of producing political content. Verifying it is another matter, and Cohen has spent two decades watching that gap widen from inside campaigns and classrooms. He has a three-part test for practitioners navigating it — real, authentic, factual — and this conversation is about why he thinks it has to be taught before anyone reaches the job. 📻 Episode overview Cohen runs Congress in Your Pocket, teaches digital campaign strategy at Johns Hopkins and NYU, and serves as executive director of Fight Hate, which works to reduce anti-Semitism on college campuses. From all of it, his argument is the same: the ethical line gets drawn before practitioners reach the job, or it does not get drawn at all. The conversation moves through what it cost him to hold a non-partisan position when one side of the political spectrum came after him, why he believes hyper-targeting served democracy better than broadcast advertising did, and what his students are starting to find they can no longer reliably spot in AI-generated video. Real, authentic, factual — he gives students that test before they touch the tools, because by the time they are on a campaign, the pressure to cross the line is already there. 🔍 Key themes discussed * What changes when AI makes political content production fast and cheap * Eighteen years of answering every user email personally — and what that reveals about civic trust * Why he teaches the ethical line before students touch the tools * Fight Hate and the deliberate choice to stop fighting hate online * What happens when AI-generated video gets good enough to fool the generation that grew up spotting it 👤 About the guest Dr. Michael Cohen lectures in political campaigning and digital strategy at Johns Hopkins University and NYU, and wrote Modern Political Campaigns: How Professionalism, Technology, and Speed Have Revolutionized Elections. He founded Congress in Your Pocket in the year of the first iPhone and has run it for eighteen years, answering every user email personally throughout. He is currently executive director of Fight Hate, working to reduce anti-Semitism on college campuses through student-led offline organising. 🕐 Chapter markers * [00:01] The iPhone as political infrastructure * [06:08] What eighteen years of personal emails taught him about trust * [13:36] Why hyper-targeting may be better for democracy than broadcast advertising * [19:31] Real, authentic, factual — the line and what it costs * [24:35] Fight Hate: using digital tools to get people off them * [37:35] The authenticity meter: how far AI video has pushed even digital natives Timestamps approximate from transcript - adjust after final edit. 🔗 Links * Dr. Michael Cohen on LinkedIn - https://www.linkedin.com/in/michaeldavidcohen/ * Congress in Your Pocket - https://www.congressinyourpocket.com * Fight Hate website - https://fighthate.org/home/ * Modern Political Campaigns (book) - https://www.modernpoliticalcampaigns.com * Blue Square Project by Robert Kraft - https://www.bluesquarealliance.org/bsa-blue-square-alliance-take-over-b/?nab=1 * Eva is on LinkedIn - https://www.linkedin.com/in/evalihotzky/

Why Security Intelligence Fails Before the Attack - Assaf Kipnis (EP 23)

Most security failures are organisational: This episode is about the gap between threat intelligence that exists and the human systems that never act on it, and what that costs the organisations that keep losing to attacks they already understood. Assaf Kipnis has spent over a decade inside the threat intelligence and trust and safety functions of some of the world's largest platforms. In this conversation, he maps a structural failure that runs across the industry: the team that identifies threats and the team that deploys detection operate in parallel, with no reliable mechanism to connect them. Intelligence gets produced, reports get written, and the knowledge sits unused while the same attacks return. Assaf describes what it actually took to stop a sophisticated actor group ahead of the 2020 US elections - a rare case where structure and resources aligned - and explains why that outcome is the exception rather than the rule. He also walks through the design decisions behind Catalyst Labs, the company he is now building to close the gap, and why he made provenance non-negotiable even at the cost of speed. 🎙 Key themes discussed * Why security teams are structurally rewarded for fighting fires rather than preventing them * The organisational gap between threat intelligence and detection - and why it persists even in well-resourced teams * What data provenance means in practice, and why it matters more than speed when using AI in security * How attackers learn your defences faster than you can adapt - and what the military analogy reveals * Why trust online currently feels, in Assaf's words, like a pipe dream 👤 About the guest Assaf Kipnis is the founder of Catalyst Labs, with over 12 years working across threat intelligence, information security, and trust and safety at LinkedIn, Google, Meta, and ElevenLabs. He brings the perspective of someone who has spent his career making threats legible to organisations - and watching those organisations lack the structure to act on what they could now see. 🕐 Chapter markers [00:18] Why the industry keeps fighting the same fires [08:04] What it actually took to stop an actor group - the 2020 elections case [12:36] How AI is widening an asymmetry that already existed [15:31] Catalyst Labs: the provenance problem and why speed comes second [20:35] What to build first if you're starting a threat intelligence team 🔗 Links Assaf Kipnis https://www.linkedin.com/in/assafkipnis/ KTLYST Labs https://www.ktlystlabs.com Background information on MGM / FBI reports: https://www.reuters.com/technology/cybersecurity/fbi-struggled-disrupt-dangerous-casino-hacking-gang-cyber-responders-say-2023-11-14/ Related episode: organisational trust and AI implementation with Simon Berkler https://open.spotify.com/episode/6y8PMaVUnZVAR1hOAR15DN Related episode: accountability and invisible infrastructure with Sergiu Petean https://open.spotify.com/episode/4KcsZBDgFzkSuwQVihjNR5