Krenar Komoni, CEO and founder of Tive, notes that agentic artificial intelligence is having a major impact on software developers, triggering supply chain's transition from recording systems to "action" systems. Komoni says that AI is beginning to "commoditize" software development. ChatGPT models like Claude from Anthropic are demonstrating the ability to write code from scratch by creating video games that would have taken people years to build. Komoni uses the phrase "It's just opened up the eyes of everybody in software development." From a supply chain perspective, Tive competes with rivals like project44, FourKites, Shippeo, Overhaul, and Roambee in the real-time visibility category, offering a multi-modal tracker hardware plus SaaS platform model. Cellular IoT, LoRaWAN, and BLE-based condition monitoring are core technologies in the cold chain and high-value goods segments.
\nIn conclusion, Komoni suggests that it may be time for some software companies to be repriced, as their value is being disrupted by AI models. For logistics and transportation companies, it is critical to begin adopting AI today. In contrast to their positions a year ago, many in that sector are now convinced that the technology works. To win with AI, however, companies will need one of three things: proprietary data; physical products or goods that cannot be replaced by AI; or direct customer relationships. From a supply chain perspective, the concept of data moat demonstrates that the defensibility of SaaS companies no longer comes from software quality or user interface, but from data ownership and network effect. As foundation models emerge, proprietary data and workflow integration depth are the two core sources of competitive advantage for vertical SaaS companies.
\nGenerative and agentic AI models are rapidly moving from recording systems to action systems. In recent months, tools like agentic voice agents have made it possible for AI to take action and be trusted by users to do so. Komoni uses the phrase "I can have a conversation with ChatGPT and feel like I'm talking to another human." This has major implications for logistics companies, which can now rely on technology to perform tasks such as issuing alarms when temperatures inside refrigerated containers exceed acceptable limits. From a supply chain perspective, cold chain monitoring is critical for compliance with USP <1079>, GDP (Good Distribution Practice), ICH Q7, and 21 CFR Part 11. Temperature excursions in the 2-8°C pharmaceutical cold chain, -25°C frozen food, and -80°C ultra-cold mRNA vaccine chain trigger product loss and recalls. Excursion management protocols are prime candidates for being delegated to the automated decision-making mechanisms of agentic AI.
\nFor now, humans remain in the loop. But ultimately, Komoni uses the phrase "they're going to start trust AI to do the right thing. Then they'll become more involved in taking care of the worst-case exceptions." From a supply chain perspective, the human-on-the-loop model, unlike human-in-the-loop, envisions humans stepping in not in every decision, but only in exceptional situations. The NIST AI Risk Management Framework and Article 14 of the EU AI Act's human oversight requirements draw the legal boundaries of this model. In operations that are not safety-critical, such as logistics, but are high-value, the pace of agentic automation adoption is higher. The API-first approach and tool use capability provide the technical infrastructure for integrating Anthropic, OpenAI, Google DeepMind, and Meta's foundation models with logistics platforms. Ultimately, Komoni's views herald a fundamental reshaping of the supply chain software market over the next 2-3 years.
\nKey Takeaways:
\n1. Komoni notes that supply chain software is transitioning from recording to action.
\n2. Anthropic's Claude can write code from scratch.
\n3. Three winning paths: proprietary data; physical product; direct customer relationship.
\n4. Refrigerated container temperature alarm is a concrete logistics application of agentic AI.
\n5. Humans will remain in exceptions; routine decisions will be delegated to AI.
[43859]
