Tech Talks Daily

What happens when the cybersecurity industry stops debating whether agentic AI is a future problem and starts treating it as a present-day reality?
In this episode of Tech Talks Daily, I sit down with Tim Freestone to unpack the biggest shift coming out of this year's RSA Conference. After attending RSA for more than two decades, Tim describes 2026 as the year the energy returned to the cybersecurity world, driven by one unavoidable topic: agentic AI.
We explore why the conversation has rapidly evolved from curiosity to urgency, and why organizations are suddenly confronting an uncomfortable truth. AI agents are already operating inside businesses, often without visibility, governance, or control. Tim explains how shadow AI is spreading faster than many leadership teams realize, with employees experimenting with autonomous tools that connect directly to company data and external AI models.
Our conversation also looks at the growing gap between visibility and control. Security teams may be discovering agents across their networks, but stopping risky behavior is an entirely different challenge. Tim argues that companies focusing purely on infrastructure are already falling behind, and that the real battleground is now the data layer itself. We discuss why data governance, audit trails, and access controls are becoming central to the future of cybersecurity strategy.
Tim also shares his thoughts on state-sponsored AI threats, the rise of autonomous espionage operations, and why open-source AI models present a completely new level of risk for defenders. At the same time, he offers practical advice for IT and security leaders trying to figure out where to start amid the noise, complexity, and endless flood of new tools entering the market.
If your organization is trying to understand how AI changes cybersecurity, governance, compliance, and risk management, this conversation offers a clear look at what security leaders are actually worried about right now, and why the next 12 months may redefine how companies think about protecting data altogether.
Useful Links
Connect with Tim Freestone

Learn More About Kiteworks

Data Security and Risk Report

Kiteworks Substack

Kiteworks LinkedIn Newsletter

Please check the partners of the Tech Tech Talks Network
Learn more about the NordLayer Browser
Visit Denodo.com

What if some of the most important applications of AI today have nothing to do with productivity, marketing, or enterprise automation, and everything to do with helping people survive crisis, displacement, and uncertainty?
In this episode, recorded at, I sit down with André Heller Pérache to explore how technology originally designed for customer service has evolved into humanitarian infrastructure supporting refugees and displaced communities around the world.
André shares the story behind Signpost, a global digital initiative from the International Rescue Committee that now operates across roughly 30 countries and 25 languages, helping register more than 20 million users while supporting over 500,000 digital social work consultations.
But this conversation goes much deeper than technology.
We discuss what happens when trusted information becomes as important as food, shelter, or medical support during times of crisis. André explains how Signpost was born from the realization that vulnerable communities were already living digitally through smartphones, WhatsApp, Facebook, and social platforms while much of the humanitarian sector still relied on traditional offline systems.
We also explore the responsible use of AI in high-stakes environments where mistakes can have real-world consequences for refugees, families, and vulnerable populations. André shares why the IRC sees AI as one of the humanitarian sector's biggest bets at a time when armed conflict, climate disasters, and shrinking budgets are putting enormous pressure on aid organizations globally.
From misinformation and trust to reducing cognitive burden and scaling empathy through technology, this episode offers a powerful reminder that behind every AI conversation are ultimately human beings searching for dignity, safety, clarity, and hope.

What if the future of AI is not one all-knowing assistant, but an entire workforce of specialized agents working together behind the scenes?
Recorded st Zendesk Relate, this episode features a fascinating conversation with Shashi Upadhyay about where enterprise AI is really heading, and why many businesses are still underestimating the scale of operational change required to make agentic AI work.
Shashi explains why Zendesk views AI agents as a new form of digital labor rather than simply another software feature. Instead of building one giant general-purpose assistant, Zendesk is developing coordinated networks of specialized agents designed for specific business functions such as billing, collections, refunds, returns, employee service, and industry-specific workflows across sectors like healthcare, banking, and e-commerce.
We also go behind the curtain inside Zendesk itself. Shashi shares how the company has transformed internally from a traditional seat-based SaaS business into an organization focused on measurable outcomes such as automation rates, customer satisfaction, and successful resolutions. He also discusses how AI is changing software development itself, enabling smaller engineering teams to move dramatically faster while reshaping how products are designed and built.
The conversation explores some of the biggest themes emerging across the AI industry right now, including outcome-based pricing, AI trust and guardrails, resolution learning loops, embedded AI, and the growing shift toward agent-to-agent interactions where personal AI assistants may eventually negotiate directly with enterprise AI systems on behalf of consumers.
We also discuss the fears many people have around jobs and automation. Rather than predicting catastrophic job loss, Shashi argues there is still enormous unmet demand for better service experiences, and that AI may ultimately allow businesses to finally deliver the level of customer experience people have wanted for years.
If you're trying to understand where enterprise AI moves next after copilots and chatbots, this conversation offers a clear and thought-provoking look at the systems, workflows, and cultural shifts already reshaping the future of work.

What if the cybersecurity industry has spent decades fighting the wrong battle?
In this episode of Tech Talks Daily, I sat down with Benny Czarny, founder and CEO of OPSWAT, to discuss why he believes the traditional "detect and respond" model is no longer enough in a world where AI is accelerating cyber threats faster than security teams can react.
Benny joined me to discuss his new book, Cybersecurity Upside Down, which combines personal stories from building OPSWAT with a bold argument for rethinking how organizations approach cyber defense altogether. His central belief is simple but provocative: detection-based security has trapped the industry in a losing cycle in which attackers need to succeed only once, while defenders are forced into a constant state of reaction.
During our conversation, Benny explained how his thinking evolved after realizing that even layering dozens of antivirus engines and sandboxing technologies still failed to stop malicious files reliably. That realization ultimately pushed him toward a prevention-first philosophy built around Deep Content Disarm and Reconstruction, or CDR. Rather than trying to determine whether a file is malicious, the approach assumes files may already be dangerous and regenerates clean, safe versions before they ever reach users or systems.
We also explored how generative AI is changing the cybersecurity landscape in ways many organizations still underestimate. Benny shared why AI is dramatically reducing the time required to create malware, weaponize exploits, and scale attacks, effectively giving even inexperienced attackers capabilities once reserved for nation states or advanced cybercriminal groups. He also raised concerns that AI data lakes could become contaminated with malicious content, creating entirely new risks for organizations rushing to deploy large language models without securing the data feeding them.
One of the most fascinating aspects of the discussion was the psychology and culture within cybersecurity teams. Benny argued that the industry often celebrates visible incident response activity while undervaluing quiet prevention. In a world dominated by alerts, dashboards, and SOC metrics, truly preventing attacks can almost appear invisible, despite potentially delivering far greater security outcomes.
We also talked about the sectors Benny believes are most exposed today, including energy, manufacturing, and critical infrastructure operators that still rely heavily on reactive security models while facing growing operational and regulatory complexity. He explained why some industries are advancing faster than others and why compliance mandates could become a major catalyst for broader prevention-first adoption.
Beyond cybersecurity itself, this episode also offered a fascinating look into Benny's entrepreneurial journey, what he learned building OPSWAT over two decades, how AI helped him research and structure his book, and why he is now even producing a cybersecurity-focused TV series called Into the Breach, designed to make complex security concepts easier for wider audiences to understand.
This conversation challenges many of the assumptions the cybersecurity industry has normalized for years. Whether you work in security, IT leadership, compliance, or want to understand how AI is reshaping digital risk, this episode offers a very different perspective on what modern cyber resilience could look like in practice.

What happens when customer service stops being a department and starts becoming an autonomous operational system?
Recorded live at, this conversation with Tom Eggemeier goes far beyond chatbots, copilots, and AI hype cycles. Instead, we explore why Zendesk believes the future of enterprise service will be built around what it calls an "autonomous service workforce," where AI agents, human experts, workflows, analytics, governance, and orchestration layers all work together as one continuously learning system.
Tom shares how Zendesk transformed its own internal operations using AI, achieving more than 60% autonomous resolution rates while simultaneously increasing customer satisfaction. We also discuss why the company is shifting away from measuring ticket deflection and toward measuring actual resolutions, what the Forethought acquisition means for Zendesk's long-term AI strategy, and why governance, permissions, and operational trust may become more important than the AI models themselves.
But this episode is about much more than software. Tom explains why he believes the next phase of enterprise AI will fundamentally reshape workflows, organizational structures, and even the role humans play inside modern businesses. We unpack the rise of specialized AI agents, why AI-to-AI interactions could soon outnumber human interactions, and why many organizations are underestimating the operational redesign required to make agentic AI work at scale.
We also discuss the hidden risks of fragmented AI systems, why disconnected tools continue to drain businesses, and how companies can balance autonomy with human oversight and empathy.
If you've been wondering where enterprise AI is really heading beyond the headlines, this conversation offers a fascinating look at how one of the biggest players in customer experience is attempting to redefine service itself.