Tech Talks Daily

Why can we track a meal traveling across town almost minute by minute, yet an international parcel can seemingly disappear between checkout and delivery?
In this episode, I speak with Dieter Van Putte from Crossborder Global, a division of Bnode that provides global commerce and logistics services. Dieter oversees technology and operations across Landmark Global and Apple Express, giving him a close view of what happens when retailers attempt to sell and deliver products across international markets.
Dieter explains why cross border delivery remains so difficult to track despite years of investment in supply chain technology. A domestic shipment may involve one main carrier, but an international order can pass between warehouses, road carriers, airports, airlines, customs authorities and final-mile delivery companies. Each participant may record events differently, operate at a different level of technical maturity or provide information at a different speed.
We discuss why the customer rarely sees this complexity. They simply expect to know when their order will arrive, what duties and taxes they must pay, plus what will happen if the product needs to be returned. When retailers cannot provide those answers, the problem quickly becomes one of trust rather than logistics alone.
Dieter also describes the less visible cost of manual reconciliation. Teams can spend hours checking carrier websites, comparing rates, reviewing spreadsheets, correcting customs information and translating inconsistent tracking events. Mistakes create further work through delayed parcels, billing disputes, customer inquiries and complaints. The direct labor cost matters, but the effect on repeat purchases and customer reviews may prove even more expensive.
Our conversation then turns to the role of automation and AI. Dieter explains how technology can support product classification, customs documentation, landed-cost calculations and exception handling. We also examine logistics control towers, which combine data from multiple parties to provide an end-to-end view of an order. With enough reliable information, these systems can detect patterns, predict delays and recommend or initiate corrective action before the customer is affected.
There is plenty of promise here, but good software cannot remove every customs rule, carrier handoff or data-quality problem. Retailers still need the right partners, accurate product information and a clear understanding of the promise they are making at checkout.
Could better use of data finally make international delivery feel as dependable as domestic shipping, and what would need to change first? Listen to the conversation and share your thoughts with me.

Do you need to understand code before you can build a technology company or lead a team of developers?
Five years after our previous conversation, I welcome Sophia Matveeva back to the podcast. Sophia is the founder of Tech for Non-Techies, where she helps founders and business professionals understand how technology products are created, tested, managed and turned into commercial ventures.
A great deal has changed since we last spoke. Generative AI tools can now turn a written description into a working prototype within hours. For someone who has spent years believing a lack of coding experience disqualified them from building a technology business, that removes a significant barrier.
Sophia believes this is the best time yet to be a non-technical founder, although her reasoning goes beyond AI-assisted coding. Research into billion-dollar technology companies shows that non-technical founders now make up a much larger share of founding teams than they did a decade ago. Many of these businesses sell technology to other companies, where commercial knowledge, customer relationships and an understanding of industry problems matter enormously.
We discuss where AI belongs in the founder journey. Sophia recommends using tools such as Lovable or Replit to create a simple test product, show it to potential customers and learn whether people would use or pay for the idea. This allows founders to test their assumptions before committing substantial money to development.
The boundary appears when that prototype becomes a real product. Once software stores customer information, processes payments or supports a commercial service, security and technical architecture cannot be treated as optional details. Sophia argues that professional developers are still needed to inspect the code, prepare the product for production and address problems a non-technical founder may not know exist.
Her point is simple. AI can help a founder reach the testing stage sooner and at a lower cost. It cannot tell someone with no engineering experience whether the generated code is safe, maintainable or ready to support paying customers.
Sophia also shares what she learned from managing her first development team. After raising investment, she attempted to compensate for her technical insecurity by taking a coding course and becoming involved in work she did not fully understand. The result was micromanagement, constant interruptions and frustrated developers.
A better approach begins with business priorities. Founders should explain what customers want, ask developers about effort and tradeoffs, agree on what will be delivered during the next work cycle, then give the team the space required to complete it. They should also allow time for technical debt, the less visible maintenance work that prevents hurried development from creating larger problems later.
We finish with advice for any business leader who wants greater technology fluency. Sophia recommends joining product meetings, contributing customer knowledge and building relationships with technical colleagues who want to understand the commercial side of the company. Neither side needs to become the other. They need enough shared language to make better decisions together.
If AI has removed the cost of testing many technology ideas, what is stopping you from finding out whether yours could work? Listen to the episode, try Sophia's exercise and share your experience with me.

What happens when an AI agent begins influencing business decisions without fully understanding the systems, processes and dependencies behind them?
In this episode, I speak with Bert van der Zwan, CEO of Bizzdesign, about the gap between enterprise AI expectations and the results many companies are seeing in practice. Bert has spent more than 25 years in software and SaaS leadership, including executive roles at Webex, Bynder, Twinfield and Unit4.
Bert offers a candid assessment of the current AI market. He believes AI will have a lasting effect on businesses and society, but argues that expectations for near-term financial returns have become inflated. Many companies are spending money on tools and experimentation without reducing costs, consolidating software or producing new revenue.
That does not mean experimentation is a mistake. Bert sees it as a necessary stage. The harder question is how companies move from a growing collection of pilots to AI capabilities that can operate dependably inside the business.
One barrier is fragmented organizational context. Large enterprises have often grown through a combination of internal expansion and acquisitions, leaving behind disconnected applications, inconsistent data definitions and processes that cross several departments. An AI system working with only part of that picture may make a fast decision, but that does not make it a good decision.
Bert argues that AI needs an authoritative view of how the enterprise works. Systems, processes, ownership, dependencies, approval status and policy restrictions must be visible and consistently defined. Without that shared context, AI may reproduce existing silos or make them worse.
We also discuss the risks boards and technology leaders should consider as AI agents become involved in operational decisions. These include unreliable data, unclear accountability, legal exposure, weak governance and an incomplete view of the process being changed. Human oversight remains necessary, particularly when an automated decision could affect customers, employees or major investments.
Bert then introduces the idea of "bespoke from the cloud." Traditional SaaS products were built around largely standardized interfaces and workflows. AI-assisted development could make software far easier to personalize around individual customers and use cases. This may give users greater control, but it could also challenge long-term software contracts and the economics that have supported the SaaS market.
For leaders trying to connect AI spending with business results, Bert recommends beginning with visibility and a clearly defined outcome. Every initiative should be judged by whether it reduces costs, increases revenue or shortens the time required to deliver value.
If AI depends on understanding how a company actually works, have businesses invested enough in creating that shared understanding before adding agents to their operations? Listen to the episode and share your thoughts with me.

Can you still trust an incoming phone call when AI can imitate a familiar voice, personalize the conversation and target information specifically to you?
In this episode, I speak with Alex Quilici, CEO of YouMail, about how artificial intelligence is changing phone fraud and why the personal devices carried by employees are becoming part of the corporate attack surface.
Alex explains how YouMail uses data from its consumer call-protection service to identify scam behavior, understand the type of fraud taking place and connect those patterns with the phone numbers involved. Advances in large language models have improved this analysis, but the same technology is also helping criminals build far more convincing campaigns.
Generic robocalls are being replaced by personalized conversations designed to extract information, impersonate trusted people and manipulate victims. Fraudsters can use AI throughout the attack chain, from identifying targets and analyzing stolen data to generating dialogue and adapting an approach during the call. Alex argues that attackers have adopted these capabilities faster than many defenders expected because successful fraud produces an immediate financial return.
The conversation also examines why voice biometrics can no longer be treated as sufficient proof of identity. As voice-cloning tools improve, companies may need to combine multiple forms of authentication and move sensitive communications into trusted applications. A call received through a banking app, for example, could give the customer greater confidence that the caller really represents their bank.
For businesses, the risk extends beyond company-managed technology. Attackers can identify where someone works, learn about their role and contact them through a personal phone that may sit outside corporate monitoring. An employee's private number can therefore provide another route into the business through impersonation and social engineering.
Alex also makes a persuasive case for collecting less personal data. Personalization can improve a service, but every additional piece of information becomes something an attacker might obtain during a breach. His advice is to identify the minimum information needed to deliver the intended experience rather than gathering data simply because it may prove useful later.
Despite the seriousness of the threat, Alex offers evidence that coordinated action can produce results. He has seen brand-impersonation campaigns reduced from tens of millions of calls each month to around 100,000 through monitoring, disruption and cooperation between businesses and telecommunications providers.
If AI is making fraudulent calls harder to recognize, should businesses stop treating the telephone network as a trusted communication channel by default? Listen to the episode and share your thoughts with me.

Why do AI agents and applications look impressive in demos but struggle when companies try to deploy them in production?
In this episode of Tech Talks Daily, I speak with Nikunj Bajaj, co-founder and CEO of TrueFoundry, about why enterprise AI has become a systems problem, what companies need to move AI from proof of concept to production, and how better infrastructure can improve reliability, governance, security, observability, and cost control.
Before founding TrueFoundry, Nikunj worked at Meta on conversational AI systems serving more than a billion users and contributed to the company's internal machine learning platforms. He explains how developers at Meta could concentrate on solving business problems while infrastructure handled logging, monitoring, deployment, and governance by default. In many enterprises, the same journey from an AI idea to a production application can still take weeks or months.
Nikunj argues that increasingly capable AI models are not necessarily the biggest barrier to enterprise adoption. The harder challenge is building reliable systems around them. Companies need to know what happens when a model becomes unavailable, how an agent is behaving, which data it can access, how much it is costing, when a human should intervene, and whether there is a kill switch when something goes wrong.
We discuss why AI proofs of concept often fail when exposed to real users. Controlled demonstrations rarely reproduce production conditions such as unexpected prompts, malicious actors, heavy workloads, model outages, latency, and dependencies between multiple components. Even when individual parts of a system perform reliably, combining them can create failure rates that businesses cannot accept for mission-critical workflows.
The conversation also examines the infrastructure required as companies introduce multiple AI models and agents. Nikunj explains the roles of model gateways, MCP gateways, and agent gateways, and how bringing these components together through an AI gateway can give enterprises a control plane for observing and governing AI traffic.
Cost is another major challenge. Nikunj explains why sending every request to the most powerful model can waste significant amounts of money when smaller or cheaper models could produce comparable results for simpler tasks. Intelligent model routing can help companies balance quality, latency, availability, and price. He shares how organizations using this approach have reduced model costs by as much as 75 to 80 percent in some production environments.
We also discuss what reliable multi-agent systems require in practice. Companies need clearly defined boundaries for what agents can do, escalation routes to other agents or people, safeguards against infinite agent loops, and complete audit trails of interactions and decisions.
For CIOs, CTOs, AI engineering teams, platform leaders, and companies trying to move generative AI and agentic AI into production, this conversation provides a practical guide to the infrastructure decisions that determine whether AI applications remain impressive prototypes or become reliable business systems.
The next stage of enterprise AI will not be defined by models alone. Companies that can connect, observe, govern, secure, and control their AI applications while managing costs will be better positioned to turn experimentation into dependable production systems.