Tech Talks Daily

In this episode of Tech Talks Daily, I sat down with Keith Zubchevich, CEO of Conviva, to unpack one of the most honest analogies I have heard about today's AI rollout.
Keith compares modern AI agents to toddlers being sent out to get a job, full of promise, curious, and energetic, yet still lacking the judgment and context required to operate safely in the real world. It is a simple metaphor, but it captures a tension many leaders are feeling as generative AI matures in theory while so many deployments stumble in practice.
As ChatGPT approaches its third birthday, the narrative suggests that GenAI has grown up. Yet Keith argues that this sense of maturity is misleading, especially inside enterprises chasing measurable returns. He explains why so many pilots stall or quietly disappoint, not because the models lack intelligence, but because organizations often release agents without clear outcomes, real-time oversight, or an understanding of how customers actually experience those interactions.
The result is AI that appears to function well internally while quietly frustrating users or failing to complete the job it was meant to do.
We also dig into the now infamous Chevrolet chatbot incident that sold a $76,000 vehicle for one dollar, using it as a lens to examine what happens when agents are left without boundaries or supervision.
Keith makes a strong case that the next chapter of enterprise AI will not be defined by ever-larger models, but by visibility. He shares why observing behavior, patterns, sentiment, and efficiency in real time matters more than chasing raw accuracy, especially once AI moves from internal workflows into customer-facing roles.
This conversation will resonate with anyone under pressure to scale AI quickly while worrying about brand risk, accountability, and trust. Keith offers a grounded view of what effective AI "parenting" looks like inside modern organizations, and why measuring the customer experience remains the most reliable signal of whether an AI system is actually growing up or simply creating new problems at speed.
As leaders rush to put agents into production, are we truly ready to guide them, or are we sending toddlers into the workforce and hoping for the best?
Useful Links
Connect with Keith Zubchevich
Learn more about Conviva
Chevrolet Dealer Chatbot Agrees to Sell Tahoe for $1
Thanks to our sponsors, Alcor, for supporting the show.

In this episode of Tech Talks Daily, I sit down with Imran Nino Eškić and Boštjan Kirm from HyperBUNKER to unpack a problem many organisations only discover in their darkest hour. Backups are supposed to be the safety net, yet in real ransomware incidents, they are often the first thing attackers dismantle. Speaking with two people who cut their teeth in data recovery labs across 50,000 real cases gave me a very different perspective on what resilience actually looks like.
They explain why so many so-called "air-gapped" or "immutable" backups still depend on identities, APIs, and network pathways that can be abused. We talk through how modern attackers patiently map environments for weeks before neutralising recovery systems, and why that shift makes true physical isolation more relevant than ever. What struck me most was how calmly they described failure scenarios that would keep most leaders awake at night.
The heart of the conversation centres on HyperBUNKER's offline vault and its spaceship-style double airlock design. Data enters through a one-way hardware channel, the network door closes, and only then is information moved into a completely cold vault with no address, no credentials, and no remote access. I also reflect on seeing the black box in person at the IT Press Tour in Athens and why it feels less like a gadget and more like a last-resort lifeline.
We finish by talking about how businesses should decide what truly belongs in that protected 10 percent of data, and why this is as much a leadership decision as an IT one. If everything vanished tomorrow, what would your company need to breathe again, and would it actually survive?
 
Useful LInks
Connect with Imran Nino Eškić
Connect With Boštjan Kirm
Learn More about HyperBUNKER
Lear more about the IT Press Tour
Thanks to our sponsors, Alcor, for supporting the show.

What happens when the AI race stops being about size and starts being about sense?
In this episode of Tech Talks Daily, I sit down with Wade Myers from MythWorx, a company operating quietly while questioning some of the loudest assumptions in artificial intelligence right now. We recorded this conversation during the noise of CES week, when headlines were full of bigger models, more parameters, and ever-growing GPU demand. But instead of chasing scale, this discussion goes in the opposite direction and asks whether brute force intelligence is already running out of road.
Wade brings a perspective shaped by years as both a founder and investor, and he explains why today's large language models are starting to collide with real-world limits around power, cost, latency, and sustainability. We talk openly about the hidden tax of GPUs, how adding more compute often feels like piling complexity onto already fragile systems, and why that approach looks increasingly shaky for enterprises dealing with technical debt, energy constraints, and long deployment cycles.
What makes this conversation especially interesting is MythWorx's belief that the next phase of AI will look less like prediction engines and more like reasoning systems. Wade walks through how their architecture is modeled closer to human learning, where intelligence is learned once and applied many times, rather than dragging around the full weight of the internet to answer every question. We explore why deterministic answers, audit trails, and explainability matter far more in areas like finance, law, medicine, and defense than clever-sounding responses.
There is also a grounded enterprise angle here. We talk about why so many organizations feel uneasy about sending proprietary data into public AI clouds, how private AI deployments are becoming a board-level concern, and why most companies cannot justify building GPU-heavy data centers just to experiment. Wade draws parallels to the early internet and smartphone app eras, reminding us that the playful phase often comes before the practical one, and that disappointment is often a signal of maturation, not failure.
We finish by looking ahead. Edge AI, small-footprint models, and architectures that reward efficiency over excess are all on the horizon, and Wade shares what MythWorx is building next, from faster model training to offline AI that can run on devices without constant connectivity. It is a conversation about restraint, reasoning, and realism at a time when hype often crowds out reflection.
So if bigger models are no longer the finish line, what should business and technology leaders actually be paying attention to next, and are we ready to rethink what intelligence really means?
Useful Links
Connect with Wade Myers
Learn More About MythWorx
Thanks to our sponsors, Alcor, for supporting the show.

What happens when we finally admit that stopping every cyberattack was never realistic in the first place?
That is the thread running through this conversation, recorded at the start of the year when reflection tends to be more honest and the noise dial is turned down a little. I was joined by returning guest Raghu Nandakumara from Illumio, nearly three years after our last discussion, to pick up a question that has aged far too well. How do organizations talk about cybersecurity value when breaches keep happening anyway?
This episode is less about shiny tools and more about uncomfortable truths. We spend time unpacking why security teams still struggle to show value, why prevention-only thinking keeps setting leaders up for disappointment, and why the conversation is slowly shifting toward resilience and containment. Raghu is refreshingly direct on why reducing cyber risk, rather than chasing impossible guarantees, is the only metric that really holds up under boardroom scrutiny.
We also talk about the strange contradiction playing out across industries. Attackers are often using familiar paths like misconfigurations, excessive permissions, and missing patches, yet many organizations still fail to close those gaps. The issue, as Raghu explains, is rarely a lack of tools. It is usually fragmented coverage, outdated processes, and a talent pipeline that blocks capable people from entering the field while claiming there is a skills shortage.
One of the most practical parts of this conversation centers on mindset. Instead of asking whether an attacker got in, Raghu argues that leaders should be asking how far they were able to go once inside. That shift alone changes how success is measured, how teams prepare for incidents, and how pressure-filled P1 moments are handled when boards want answers every fifteen minutes.
We also touch on how legal action, public claims campaigns, and customer lawsuits are changing the stakes after a breach, forcing executives to rethink how they frame cyber investment. From there, Raghu shares how Illumio has been working with Microsoft to strengthen internal resilience at massive scale, and why visibility and segmentation are becoming harder to ignore.
This is a conversation about realism, responsibility, and growing up as an industry. If cybersecurity is really about safety and not slogans, what would you want your organization to stop saying, and what would you rather hear instead?
Please feel free to upload the podcast. Here are also the links we discussed on the call:
Useful Links
Connect with Raghu Nandakumara on LinkedIn and Twitter
Learn more about Illumio
Lateral Movement in Cyberattacks
Illumio Podcast 
Follow on Facebook, Twitter, LinkedIn, and YouTube
 
Thanks to our sponsors, Alcor, for supporting the show.

What really happens inside an organization when a cyber incident hits and the neat incident response plan starts to fall apart?
That question sat at the heart of my return conversation with Max Vetter, VP of Cyber at Immersive. It has been a big year for breaches, public fallout, and eye-watering financial losses, and this episode goes beyond headlines to examine what cyber crisis management actually looks like when pressure, uncertainty, and human behavior collide. Max brings a rare perspective shaped by years in law enforcement, intelligence work, and hands-on cyber defense, and he is refreshingly honest about where most organizations are still unprepared.
We talked about why written incident response plans tend to fail at the exact moment they are needed most. Cyber incidents are chaotic, emotional, and non-linear, yet many plans assume calm decision-making and perfect coordination. Max explains why success or failure is often defined by the response rather than the initial breach itself, and why leadership, communication, and judgment matter just as much as technical skill. Real-world examples from major incidents highlight how competing pressures quickly emerge, whether to contain or keep systems running, whether to pay a ransom or risk prolonged downtime, and how every option comes with consequences.
One idea that really stood out is Max's belief that resilience is revealed, not documented. Compliance and audits may tick boxes, but they rarely expose how teams behave under stress. We explored why organizations that rely on annual tabletop exercises often develop a false sense of confidence, and how that confidence can become dangerous when decisions are made quickly and publicly. Max shared why the best-performing teams are often the ones that feel less certain in the moment, because they question assumptions and adapt faster.
We also dug into the growing role of crisis simulations and micro-drills. Rather than rehearsing a single scenario once a year, Immersive focuses on repeated, realistic practice that builds muscle memory across technical teams, executives, legal, and communications. The goal is not to predict the exact attack, but to train people to think clearly, collaborate across functions, and make defensible decisions when there are no good options. That preparation becomes even more important as cyber incidents increasingly spill into supply chains, manufacturing, and the physical world.
As public scrutiny rises and consumer-led legal action becomes more common after breaches, reputation and response speed now sit alongside forensics and recovery as business-critical concerns. This episode is a candid look at why cyber crisis readiness is a discipline, not a document, and why assuming you will cope when the moment arrives is a risky bet.
So if resilience only truly shows itself when everything is on the line, how confident are you that your organization would perform when the pressure is real and the clock is ticking?
Useful Links
Connect with Max Vetter on Linkedin
Learn more about Immersive Labs
Follow on LinkedIn, Instagram, Twitter and Facebook

Thanks to our sponsors, Alcor, for supporting the show.