DX Today | No-Hype Podcast & News About AI & DX

This episode of DX Today, recorded January 17, 2026, examines the definitive market shift from generative AI to agentic AI—where systems move from answering questions to autonomously executing tasks. Hosts Alex and Mike break down the distinct strategies of the major players: OpenAI's "Operator" feature now handles complete transactions like booking flights without user intervention, Google's Gemini 3 generates custom interactive interfaces on demand rather than text responses, xAI's Grok 4.1 has pivoted toward emotional intelligence for creative and customer service applications, and Meta is winning the "physical world" with AI wearables and neural control bands that enable ambient, screen-free computing.

Beyond the technology, the episode addresses the practical realities facing enterprise leaders: power infrastructure is becoming the primary bottleneck for scaling agentic systems, the regulatory landscape remains fragmented with the EU AI Act's high-risk provisions hitting in August 2026 while the US lacks unified federal standards, and the workforce conversation has shifted from job replacement fears to "AI literacy" as a baseline requirement—with "Agent Manager" emerging as the critical skill set. The through-line reinforces DX Today's editorial position: the novelty phase is over, and competitive advantage now comes not from having the smartest model but from building the best workflows. Facts, no hype.

Send us a text

Agentic AI represents a fundamental paradigm shift in cybersecurity, moving beyond pattern analysis to autonomous decision-making and action. Unlike generative AI, these systems can execute complex, multi-step tasks with minimal human oversight, fundamentally restructuring both defensive and offensive cyber operations. Gartner has named agentic AI the top technology trend for 2025, projecting that 33% of enterprise applications will incorporate it by 2028.

For defenders, agentic AI offers significant operational value by automating Security Operations Center (SOC) workflows, enabling continuous vulnerability testing, executing incident response at machine speed, and performing sophisticated behavioral threat analysis. Early adopters report substantial reductions in incident detection and response times. However, implementation is fraught with challenges, including the probabilistic and non-deterministic nature of AI, the proliferation of difficult-to-manage non-human identities (NHIs), a lack of transparency in AI decision-making, and significant supply chain vulnerabilities.

The same capabilities are being weaponized by adversaries with devastating effect. In September 2025, the first large-scale cyberattack executed with minimal human intervention was documented, in which Chinese state-sponsored actors used agentic AI for espionage. This event signals that the barrier to entry for sophisticated attacks has dropped significantly. Offensive techniques now include memory poisoning, tool misuse via prompt injection, multi-agent compromises, and deepfake-enabled social engineering at scale, as evidenced by a $25 million fraud against the engineering firm Arup.

This is not a future threat; it is a current operational reality. Business leaders must treat agentic AI not as another tool but as a strategic imperative, addressing the complex questions of governance, validation, and security for these autonomous systems. Disciplined implementation that prioritizes transparency, robust identity management, and security from the outset is critical to realizing the benefits while mitigating catastrophic risks.

Send us a text

 The paradigm shift from Generative AI to Agentic AI, marking a transition from digital tools that merely suggest products to autonomous "Actionbots" that execute transactions. These sophisticated agents utilize perception, reasoning, and memory to navigate the web, negotiate prices, and manage complex workflows like travel planning on behalf of consumers. As the primary customer evolves from a human to an algorithm, retailers must move beyond traditional SEO toward Agent Optimization by prioritizing machine-readable APIs and structured data. The text highlights major industry players like OpenAI’s Operator and Amazon Rufus, noting that businesses must adapt to this automated economy or face obsolescence. While this evolution promises immense efficiency, it also introduces significant challenges regarding consumer trust, data privacy, and regulatory compliance. Ultimately, the document serves as a strategic roadmap for navigating a future where the friction between consumer desire and fulfillment is virtually eliminated.

Send us a text

In early 2026, the AI landscape shifted from simple "Chat" and "Retrieval Augmented Generation" (RAG) to Deep Research Agents—systems capable of autonomous, multi-day investigations, cross-document synthesis, and complex reasoning. However, a critical bottleneck emerged: How do you evaluate an AI that knows more than the evaluator?

Traditional benchmarks (static Q&A pairs) fail to capture the nuance of a 50-page due diligence report or a legal discovery synthesis. Enter the era of Deep Research Evaluation, an emerging field of frameworks currently trending among AI researchers. This paper proposes a paradigm shift: using Agentic Evaluation to test Agentic AI.

These new evaluation methodologies introduce fully automated pipelines that generate complex, persona-based research tasks and evaluate the results using dynamic, adaptive criteria and active fact-checking—even when citations are missing. Early industry observations of leading systems like Gemini 2.5 Pro and OpenAI Deep Research reveal that while reasoning has improved, "hallucination in synthesis" remains a critical enterprise risk.

This report analyzes the landscape of deep research evaluation frameworks, their market implications, and provides a roadmap for enterprises to adopt "Agentic Testing" for their most complex AI workflows.

The critical distinction between legitimate synthetic media and malicious deepfakes, arguing against the societal impulse to "kill the messenger" by demonizing the technology itself. While deepfakes are weaponized for fraud and identity theft, the source highlights how synthetic delivery offers immense benefits for journalism, education, and accessibility through scalable, data-driven avatars. The author identifies a "crisis of categorization" where low-quality "AI slop" and deceptive content prejudice the public against helpful innovations like digital twin meteorologists or sign language interpreters. To preserve the utility of these tools, the text advocates for a "Truth Layer" using technical standards like C2PA to provide transparent content credentials. Ultimately, the passage suggests that society must move toward verifying the provenance of digital messengers rather than stifling the technological medium.