Formerly known as Wikibon
Book a Briefing

How To Build Decision-grade AI Agents You Can Trust and Audit

Next Frontiers of AI podcast on decision-grade AI agents.

Enterprises are pushing agentic AI beyond copilots into diagnosis, problem-solving, and decision-making—but trust is now the ROI limiter. In this episode of Next Frontiers of AI, Scott Hebner and George Gilbert explain why LLM-only architectures are reliability traps and outline a practical, three-layer blueprint—LLM+CoT, semantic layers (knowledge graphs), and causal reasoning—to deliver decisions you can verify, defend, and audit.

303 | Breaking Analysis | Enterprise Technology Predictions 2026

At the beginning of each year, as is our tradition, we team up with ETR to dig through the latest data and craft ten predictions for the coming year. This year’s prognostication follows the publication where we grade our 2025 predictions. In this Breaking Analysis, we tap some of the most telling nuggets from ETR’s rich data set and put forth our top ten predictions for enterprise tech in 2025.

302 | Breaking Analysis | 2026 Data Predictions: Scaling Agents via Contextual Intelligence

More than eight years into the modern era of AI, the industry has moved past the awe of GenAI 1.0. The novelty is gone and in its place is scrutiny. Enterprises are less impressed by demos and more impatient about outcomes. Market watchers are increasingly skeptical about vague narratives; and the commentary has shifted from “look what AI can do” to “show me the money, give me visibility and control.”
This dynamic is backstopped by our central premise that the linchpin of AI is data; and specifically data in context. The modern data stack of the 2010s – i.e. cloud-centric, separation of compute and storage, pipelines, dashboards, etc.- now feels trivial. The target has moved to enabling agentic systems that can act, coordinate, and learn across enterprises with a mess of structured and unstructured data, complex workflows, conflicting policies, multiple identities, and the nuanced semantics that live inside business processes.
The industry remains excited and at the same time conflicted. To tap an oft-cited baseball analogy – the first inning was academic discovery in and around 2017 – papers explaining transformers and diffusion models. But most people weren’t paying attention. The second inning was the ChatGPT moment and the AI heard ‘round the world, which has brought excitement and plenty of hype.

Will 2026 Be The Year AI Decision Intelligence Goes Mainstream? 

In this episode of Next Frontiers of AI, Scott Hebner and Joel Sherlock, CEO of Causify, argue that 2026 will be the year AI Decision Intelligence goes mainstream. Following GenAI and the rise of AI agents and agentic workflows, enterprises are facing a reality check, as a recent Carnegie Mellon study found — AI agents can act, but they often cannot justify, explain, or audit the decisions that matter most. Scott and Joel unpack why causal AI and knowledge graphs are emerging as the enabling layer for decision-grade AI.

From Digital to Physical: How End-User-Centered AI Is Transforming Industrial Work

Futuristic scene showing an Oshkosh autonomous refuse truck, airport security robot, and construction worker using AI systems at CES 2026, representing physical AI for end users.

At CES 2026, Oshkosh revealed how physical AI is moving artificial intelligence out of the cloud and into the real world, powering refuse trucks, airport security robots, and job-site systems that directly support frontline workers. By combining edge AI, robotics, and human-in-the-loop autonomy, Oshkosh is showing how end-user-centered AI can deliver safer, more productive, and more trusted operations across fleets, airports, and industrial environments.

Book A Briefing

Fill out the form , and our team will be in touch shortly.
Skip to content