JUXT AI Radar
Introduction
2025 delivered on the promise of agentic AI. After years of bold predictions, we’re now seeing genuine production value, and the impact is profound.
For software engineering, this represents the most significant transition since compilers changed what it meant to write code, but the implications reach even further. Agentic AI provides a co-intelligence that puts capabilities once requiring technical specialists within everyone’s reach, much as spreadsheets did a generation ago: transformative business value and new categories of risk to manage. We’re still early in understanding the full scope of this change, and this quarter’s radar captures both the areas that are maturing and the emerging frontiers that point to what comes next.
On the maturity side, we’ve added Temporal to our Adopt ring, recognising that durable workflow orchestration has become essential infrastructure for agents that need to survive failures and run reliably over days rather than seconds. Organisations are moving beyond prototypes to ask harder questions: how do we understand our processes before automating them? How do we monitor agents in production? The new entries for process mining and LLM observability reflect this shift from experimentation to operation.
On the frontier side, we’re seeing growing interest in approaches that extend beyond what LLMs alone can achieve. Ontologies provide grounded semantics, giving AI systems authoritative definitions and relationships rather than statistical associations. Neurosymbolic AI couples neural networks with symbolic reasoning, enabling explainable decisions and guaranteed rule compliance. World models build internal representations that simulate how environments behave, enabling systems to predict consequences rather than merely paraphrase scenarios from training data. Together, these approaches point toward AI architectures that combine LLM flexibility with the formal semantics that regulated industries require.
AI is also escaping the screen. Foundation models purpose-built for robotics and physical systems are maturing rapidly, and digital twin platforms such as NVIDIA Omniverse are enabling organisations to train and test AI in simulation before deployment. We’ve added coverage of both areas for teams working at the intersection of AI and the physical world.
As we enter 2026, this radar captures where the landscape stands today. We’re still in the early stages of this transformation, one we’re following with professional and personal interest.
— Henry Garner (CTO, JUXT), January 2026
Radar Overview
Our radar is organized into four main categories, each containing technologies evaluated across four adoption levels:
- Adopt: Technologies we recommend using now
- Trial: Worth exploring for new projects
- Assess: Keep under observation
- Hold: Not recommended for new projects
Categories
Techniques
AI methodologies and practices that shape how we build intelligent systems.
Languages & Frameworks
Programming languages and frameworks that power AI development.
Tools
Software tools and utilities that enhance AI development workflows.
Platforms
Infrastructure and platform services that support AI applications.
Contributing
This radar represents our current viewpoint and will be updated regularly. We welcome feedback and suggestions from the community, you can reach us on LinkedIn, BlueSky and via email. Each technology entry includes detailed reasoning for its placement, helping you make informed decisions for your AI projects.
