3534: Agentic AI at Scale: What 120 Million Monthly Conversations Really Mean

What does it really mean when AI moves from answering questions to making decisions that affect real people, real money, and real outcomes? In this episode of Tech Talks Daily, I'm joined by Joe Kim, CEO of Druid AI, for a grounded conversation about why agentic AI is becoming the focus for enterprises that have moved beyond experimentation. After years of hype around generative tools, many organizations are now facing a tougher question. Can AI be trusted to take action inside core business processes, and can it do so with the accuracy, security, and accountability that enterprises expect? Joe brings a rare perspective shaped by decades leading large-scale enterprise software companies, including his time as CEO of Sumo Logic. He explains why Druid AI deliberately avoids positioning itself as a generative AI company, and instead focuses on systems that can make decisions, trigger workflows, and complete tasks inside regulated, high-stakes environments. We unpack why accuracy thresholds matter when AI touches billing, healthcare, admissions, or compliance, and why security and governance are no longer secondary concerns once AI is allowed to act. We also talk about scale and proof. Druid AI now supports over 120 million conversations every month, a figure that keeps climbing as enterprises move agentic systems into production. Joe shares how those conversations translate into measurable business outcomes, from operational efficiency to revenue growth, and why many AI initiatives fail to reach this stage. His "5 percent club" philosophy cuts through the noise, focusing on the small number of use cases that actually deliver return while most others stall in pilots. The conversation also explores why higher education has become a surprising pressure point for AI adoption, how outdated systems contribute to student churn, and how conversational agents can remove friction at moments that decide whether someone enrolls, stays, or leaves. We close by looking ahead at Druid AI's next chapter, including new platform capabilities designed to make building and deploying agents faster without sacrificing control. As more enterprises demand results instead of promises, are we ready to judge AI by the decisions it makes and the outcomes it delivers, and what should that accountability look like in your organization? I'd love to hear your thoughts. Where do you see agentic AI delivering real value today, and where do you think the risks still outweigh the rewards? What does it really mean when AI moves from answering questions to making decisions that affect real people, real money, and real outcomes? In this episode of Tech Talks Daily, I'm joined by Joe Kim, CEO of Druid AI, for a grounded conversation about why agentic AI is becoming the focus for enterprises that have moved beyond experimentation. After years of hype around generative tools, many organizations are now facing a tougher question. Can AI be trusted to take action inside core business processes, and can it do so with the accuracy, security, and accountability that enterprises expect? Joe brings a rare perspective shaped by decades leading large-scale enterprise software companies, including his time as CEO of Sumo Logic. He explains why Druid AI deliberately avoids positioning itself as a generative AI company, and instead focuses on systems that can make decisions, trigger workflows, and complete tasks inside regulated, high-stakes environments. We unpack why accuracy thresholds matter when AI touches billing, healthcare, admissions, or compliance, and why security and governance are no longer secondary concerns once AI is allowed to act. We also talk about scale and proof. Druid AI now supports over 120 million conversations every month, a figure that keeps climbing as enterprises move agentic systems into production. Joe shares how those conversations translate into measurable business outcomes, from operational efficiency to revenue growth, and why many AI initiatives fail to reach this stage. His "5 percent club" philosophy cuts through the noise, focusing on the small number of use cases that actually deliver return while most others stall in pilots. The conversation also explores why higher education has become a surprising pressure point for AI adoption, how outdated systems contribute to student churn, and how conversational agents can remove friction at moments that decide whether someone enrolls, stays, or leaves. We close by looking ahead at Druid AI's next chapter, including new platform capabilities designed to make building and deploying agents faster without sacrificing control. As more enterprises demand results instead of promises, are we ready to judge AI by the decisions it makes and the outcomes it delivers, and what should that accountability look like in your organization? I'd love to hear your thoughts. Where do you see agentic AI delivering real value today, and where do you think the risks still outweigh the rewards? Useful Links Connect with Joe Kim, CEO of Druid AI. Druid AI Website Tech Talks Daily is Sponsored by Denodo  

The world is building data centers, identity rails, and AI policy stacks at a speed that makes 2026 feel closer than it is. In this conversation, Rajesh Natarajan, Global Chief Technology Officer at Gorilla Technology Group, explains what it takes to engineer platforms that remain reliable, secure, and sovereign-ready for decades, especially when infrastructure must operate outside the safety net of constant cloud connectivity. Raj talks about quantum-safe networking as a current risk, not a future headline. Adversaries are capturing encrypted traffic today, betting on decrypting it later, and retrofitting quantum-safe architecture into national platforms mid-lifecycle is an expensive mistake waiting to happen. He also highlights the regional nature of AI infrastructure, Southeast Asia prioritizing sovereignty, speed, and efficiency, Europe leaning on regulation and telemetry, and the U.S. betting on raw cluster scale and throughput. Sustainability at Gorilla isn't a marketing headline, it's an engineering requirement. If a system can't prove its environmental impact using telemetry like workload-level PUE, it isn't labeled sustainable internally. Gorilla applies the same rigor to IoT insight per unit of energy, device lifecycles, and edge-level intelligence placement, minimizing data centralization without operational justification. This episode offers marketers, founders, and technology leaders a rare chance to understand what national-scale resilience looks like when platform alignment breaks first, not technology. Remembering that decisions must be reversible, explicit, and measurable is the foundation of how Gorilla is designing systems that can evolve without forcing rushed compromises when uncertainty becomes reality. Useful links: Connect with Dr Rajesh Natarajan Gorilla website Tech Talks Daily is Sponsored by Denodo  

What makes live events feel personal in an age of algorithms making the calls? That's the tension marketers are living in right now. Ben Kruger, Chief Marketing Officer at Event Tickets Center, sits at the center of this shift. He has spent 20 years shaping server-side systems and performance marketing strategies, including a decade of persistence chasing a role at Google before landing a position in New York just as eCommerce demand went into overdrive during the pandemic. Now, at ETC, he runs marketing for more than 130,000 live events simultaneously. It's a scale that forces automation to step in. The industry moves in real time, resellers update prices by the hour, artists trend globally overnight, weather can shift demand before a stadium gate opens. Ben credits Google's AI tools and internal models as a competitive advantage, but he also talks openly about the risks. The early excitement of automation gave way to skepticism after seeing unaligned promises from new platforms and unpredictable campaign behavior in tools that remove control from brands. There's a well-rounded argument to explore here. On one side, AI enables a small team to do the work of thousands, writing content at a volume no human team could deliver alone. On the other, removing risk from campaigns, or removing channel-level choices from advertisers, can reduce trust and increase low-quality creative output. Advantage+ tools that make placement decisions automatically, without brand input, might scale reach, but can reduce clarity of intent and control of outcomes. Some CMOs see that as smart acceleration, others see it as an overcorrection that creates opacity and dependency on platforms optimizing for their own incentives. And somewhere in the middle is the opportunity. ETC's approach shows a future where repetition in rapid testing generates sharper insight, where lean teams move faster, where humans stay in the loop to validate outcomes, and where creativity stays grounded in audience understanding, economics, and transparency. Marketers listening to Ben will hear someone who wants experimentation, control, clarity, and long-term audience trust to exist side by side. Useful links: Connect with Ben Kruger on LinkedIn  Event Tickets Center website Tech Talks Daily is Sponsored by Denodo

What does it really take to build software that can grow from a single line of code to millions of users a day without losing its soul along the way? In this episode of Tech Talks Daily, I'm joined by Alex Gusev, CTO at Uploadcare, for a wide-ranging conversation about scale, simplicity, and why leadership in technology starts with people long before it gets anywhere near frameworks or tooling. Alex has spent two decades building server-side systems, often inside small teams, and has seen firsthand how early decisions echo through a company's future, for better and for worse. We talk openly about the realities of early-stage engineering, including why shipping imperfect code is often the only way to survive, how technical debt should be taken on deliberately rather than by accident, and why knowing when to slow down and clean things up is one of the hardest leadership calls to make. Alex shares his belief that simplicity is the strongest ally in high-load environments, and how over-engineering, often inspired by copying the playbooks of much larger companies, creates fragility instead of strength. Our conversation also digs into his continued faith in Ruby on Rails, a framework that divides opinion but still plays a central role in many successful products. Alex reframes the debate around speed, focusing less on raw performance metrics and more on how quickly teams can build, adapt, and maintain systems over time. It's a practical view shaped by real-world trade-offs rather than theory. Beyond code, we explore why Alex puts people ahead of technology and process, and how creating psychological safety inside teams leads to better decisions, lower churn, and smarter use of limited resources. He also reflects on personal experiences that reshaped his approach to leadership, the growing tech scene in Kyrgyzstan, and why he finds as much inspiration in Dostoevsky as he does in engineering blogs. If you've ever questioned whether modern engineering culture has overcomplicated itself, or wondered how to balance ambition with sustainability as your product grows, this episode offers plenty to think about. Where do you think your own team is adding complexity without realizing it, and what might change if you started with people first? Useful Links Connect with Alex Gusev Learn more about Uploadcare Tech Talks Daily is sponsored by Denodo

If you have ever opened Candy Crush over the holidays without thinking about the design decisions behind every swipe, this episode offers a rare look behind the curtain.  I sit down with Abigail Rindo, Head of Creative at King, to unpack how accessibility has evolved from a well-meaning afterthought into a core creative and commercial practice inside one of the world's most recognizable gaming studios.  With more than 200 million people playing King's games each month, Abigail explains why inclusive design cannot be treated as charity or compliance, but as a responsibility that directly shapes product quality, player loyalty, and long-term growth. One of the moments that really stayed with me in this conversation is the data. More than a quarter of King's global player base self identifies as having an accessibility need. Even more players benefit from accessibility features without ever labeling themselves that way. Abigail shares how adjustments like customizable audio for tinnitus, reduced flashing to limit eye strain, and subtle interaction changes can quietly transform everyday play for millions of people. These are not edge cases. They are everyday realities for a massive audience that lives with these games as part of their daily routine. We also dig into how inclusive design sparks better creativity rather than limiting it. Abigail walks me through updates to Candy Crush Soda Saga, including the "hold and drag" mechanic that allows players to preview a move before committing. Inspired by the logic of holding a chess piece before placing it, this feature emerged directly from player research around visibility, dexterity, and comfort. It is a reminder that creative constraints, when grounded in real human needs, often lead to smarter and more elegant solutions. Beyond mechanics and metrics, this conversation goes deeper into storytelling, empathy, and team culture. Abigail explains why inclusive design only works when inclusive teams are involved from the start, and how global storytelling choices help King design worlds that resonate everywhere from Stockholm to Antarctica. We also talk about live service realities, blending quantitative data about what players do with qualitative insight into why they do it, especially when a game has been evolving for more than a decade.  

What does it actually mean to prove who we are online in 2025, and why does it still feel so fragile? In this episode of Tech Talks Daily, I sit down with Alex Laurie from Ping Identity to talk about why digital identity has reached a real moment of tension in the UK. As more of our lives move online, from banking and healthcare to social platforms and government services, the gap between how identity should work and how it actually works keeps widening. Alex shares why the UK now feels out of step with other regions when it comes to online identity schemes, and how heavy reliance on centralized models is slowing adoption while weakening public trust. We spend time unpacking the practical consequences of today's verification systems. Age checks are regularly bypassed, fraud continues to grow, and users are often asked to hand over far more personal data than feels reasonable just to access everyday services. At the same time, public pressure around online safety is rising fast. That creates an uncomfortable push and pull between tighter controls and the expectation of fast, low-friction access.  Alex makes the case that this tension exists because the underlying approach is flawed, and that proving something simple, like age, should never require revealing an entire digital identity. From there, the conversation turns to decentralized identity and why it is gaining momentum globally. Instead of placing sensitive data into large centralized databases, decentralized models allow individuals to hold and present verified credentials on their own terms. For me, this reframes digital identity as a right rather than a feature, and opens the door to systems that feel more privacy-aware, inclusive, and resilient. We also explore how agentic AI could play a role here, helping people manage, present, and protect their credentials intelligently without adding complexity or new risks. With fresh consumer research from Ping Identity informing the discussion, this episode looks closely at where trust, privacy, and identity are heading next, and why the choices made now will shape how we prove who we are online for years to come. Are we finally ready to rethink digital identity, and if so, what does that mean for all of us?

What does it really mean to keep humans at the center of AI when agentic systems are accelerating faster than most organizations can govern them? At AWS re:Invent, I sat down with Michael Bachman from Boomi for a wide-ranging conversation that cut through the hype and focused on the harder questions many leaders are quietly asking. Michael leads technical and market research at Boomi, spending his time looking five to ten years ahead and translating future signals into decisions companies need to make today. That long view shaped a thoughtful discussion on human-centric AI, trust versus autonomy, and why governance can no longer be treated as an afterthought. As businesses rush toward agentic AI, swarms of autonomous systems, and large-scale automation, Michael shared why this moment makes him both optimistic and cautious. He explained why security, legal, and governance teams must be involved early, not retrofitted later, and why observability and sovereignty will become non-negotiable as agents move from experimentation into production. With tens of thousands of agents already deployed through Boomi, the stakes are rising quickly, and organizations that ignore guardrails today may struggle to regain control tomorrow. We also explored one of the biggest paradoxes of the AI era. The more capable these systems become, the more important human judgment and critical thinking are. Michael unpacked what it means to stay in the loop or on the loop, how trust in agentic systems should scale gradually, and why replacing human workers outright is often a short-term mindset that creates long-term risk. Instead, he argued that the real opportunity lies in amplifying human capability, enabling smaller teams to achieve outcomes that were previously out of reach. Looking further ahead, the conversation turned to the limits of large language models, the likelihood of an AI research reset, and why future breakthroughs may come from hybrid approaches that combine probabilistic models, symbolic reasoning, and new hardware architectures. Michael also reflected on how AI is changing how we search, learn, and think, and why fact-checking, creativity, and cognitive discipline matter more than ever as AI assistants become embedded in daily life. This episode offers a grounded, future-facing perspective on where AI is heading, why integration platforms are becoming connective tissue for modern systems, and how leaders can approach the next few years with both ambition and responsibility.  Useful Links Learn More About Boomi Connect with Michael Bachman Algorithms to Live By: The Computer Science of Human Decisions Tech Talks Daily is sponsored by Denodo

What does responsible AI really look like when it moves beyond policy papers and starts shaping who gets to build, create, and lead in the next phase of the digital economy? In this conversation recorded during AWS re:Invent, I'm joined by Diya Wynn, Principal for Responsible AI and Global AI Public Policy at Amazon Web Services. With more than 25 years of experience spanning the internet, e-commerce, mobile, cloud, and artificial intelligence, Diya brings a grounded and deeply human perspective to a topic that is often reduced to technical debates or regulatory headlines. Our discussion centers on trust as the real foundation for AI adoption. Diya explains why responsible AI is not about slowing innovation, but about making sure innovation reaches more people in meaningful ways. We talk about how standards and legislation can shape better outcomes when they are informed by real-world capabilities, and why education and skills development will matter just as much as model performance in the years ahead. We also explore how generative AI is changing access for underrepresented founders and creators. Drawing on examples from AWS programs, including work with accelerators, community organizations, and educational partners, Diya shares how tools like Amazon Bedrock and Amazon Q are lowering technical barriers so ideas can move faster from concept to execution. The conversation touches on why access without trust falls short, and why transparency, fairness, and diverse perspectives have to be part of how AI systems are designed and deployed. There's an honest look at the tension many leaders feel right now. AI promises efficiency and scale, but it also raises valid concerns around bias, accountability, and long-term impact. Diya doesn't shy away from those concerns. Instead, she explains how responsible AI practices inside AWS aim to address them through testing, documentation, and people-centered design, while still giving organizations the confidence to move forward. This episode is as much about the future of work and opportunity as it is about technology. It asks who gets to participate, who gets to benefit, and how today's decisions will shape tomorrow's innovation economy. As generative AI becomes part of everyday business life, how do we make sure responsibility, access, and trust grow alongside it, and what role do we each play in shaping that future? Useful Links Connect With Diya Wynn AWS Responsible AI Tech Talks Daily is sponsored by Denodo

What does it really mean to support developers in a world where the tools are getting smarter, the expectations are higher, and the human side of technology is easier to forget? In this episode of Tech Talks Daily, I sit down with Frédéric Harper, Senior Developer Relations Manager at TinyMCE, for a thoughtful conversation about what it takes to serve developer communities with credibility, empathy, and long-term intent. With more than twenty years in the tech industry, Fred's career spans hands-on web development, open source advocacy, and senior DevRel roles at companies including Microsoft, Mozilla, Fitbit, and npm. That journey gives him a rare perspective on how developer needs have evolved, and where companies still get it wrong. We explore how starting out as a full-time developer shaped Fred's approach to advocacy, grounding his work in real-world frustration rather than abstract messaging. He reflects on earning trust during challenging periods, including advocating for open source during an era when some communities viewed large tech companies with deep skepticism. Along the way, Fred shares how studying Buddhist philosophy has influenced how he shows up for developers today, helping him keep ego in check and focus on service rather than status. The conversation also lifts the curtain on rich text editing, a capability most users take for granted but one that hides deep technical complexity. Fred explains why building a modern editing experience involves far more than formatting text, touching on collaboration, accessibility, security, and the growing expectations around AI-assisted workflows. It is a reminder that some of the most familiar parts of the web are also among the hardest to build well. We then turn to developer relations itself, a role that is often misunderstood or measured through the wrong lens. Fred shares why DevRel should never be treated as a short-term sales function, how trust and community take time, and why authenticity matters more than volume. From open source responsibility to personal branding for developers, including lessons from his book published with Apress, Fred offers grounded advice on visibility, communication, and staying human in an increasingly automated industry. As the episode closes, we reflect on burnout, boundaries, and inclusion, and why healthier communities lead to better products. For anyone building developer tools, managing technical communities, or trying to grow a career without losing themselves in the process, this conversation leaves a simple question hanging in the air: how do we build technology that supports people without forgetting the people behind the code? Useful Links Connect with Frédéric Harper Learn More About TinyMCE Tech Talks Daily is sponsored by Denodo

What does it really take to build a fintech company that quietly fixes one of the most frustrating problems SMEs face every day? In this episode of Tech Talks Daily, I'm joined by Pierre-Antoine Dusoulier, the Founder and CEO of iBanFirst, for a candid conversation about entrepreneurship, timing, and why cross-border payments have remained broken for so long.  Pierre-Antoine's story begins in London, where his early career as an FX trader felt like a compromise at the time, yet quietly gave him a front-row seat to inefficiencies most people accepted as normal. That experience would later shape two companies and a very clear point of view on how money should move across borders. Pierre-Antoine walks through his first venture, Combeast.com, one of France's earliest FX brokerages for retail investors, and what he learned from selling it to Saxo Bank and staying on to run Western European operations. That chapter matters, because it exposed the gap between how sophisticated FX markets really are and how poorly SMEs are served when FX and payments are bundled together inside traditional banks. Out of that frustration, IbanFirst was born in 2016 with a simple idea: treat cross-border payments as a specialist discipline, not a side feature. Today, IbanFirst serves more than 10,000 clients across Europe and processes over €2 billion in transactions every month. We dig into why growth has continued while many fintechs have slowed, from a product designed to be used daily, to proactive sales, to a new generation of CFOs and CEOs who expect the same clarity and speed at work that they get from consumer fintech tools.  Pierre-Antoine explains how real-time FX rates, payment tracking using SWIFT GPI, and multi-entity account management change the day-to-day reality for SMEs trading internationally. We also talk about Brexit, and how being rooted in continental Europe created an unexpected opening. Pierre-Antoine shares why expanding into the UK, including the acquisition of Cornhill, made sense, and why London's payments ecosystem still stands apart in scale and depth. Along the way, he is refreshingly open about the heavy investment required in compliance, trust, and regulation, and why nearly a third of IbanFirst's team focuses on operations and oversight. Looking ahead, Pierre-Antoine lays out a bold vision for the SME payments market, predicting a future where specialists replace banks in much the same way fintech reshaped consumer money transfers. As cross-border trade grows and currency volatility becomes a daily concern, his perspective raises an interesting question for anyone running an international business today:  if specialists already exist, why keep relying on systems that were never designed for how SMEs actually operate? Useful Links: Connect with Pierre-Antoine Dusoulier Learn more about iBanFirst, Tech Talks Daily is sponsored by Denodo