JUXT AI Radar
Introduction
Frontier models and AI coding assistants have continued to improve at a relentless pace since our last update, and many software engineering teams are now working almost entirely within an agentic coding harness. In our work on agentic AI platforms for large enterprises, we see first-hand the same transition being rolled out to other knowledge workers.
What limits adoption now is the engineering and security work needed to turn their capabilities into reliable human-AI “centaur” systems. Most of this quarter’s updates are in this area.
In the Adopt ring, MCP and RAG picked up substantial new security guidance covering authentication, least-privilege tool grants, audit logging and defences against indirect prompt injection. CaMeL enters our radar in Assess as a research-level answer to the same problem. There have been some very public failures this quarter, from Claude Code’s own source code being accidentally exposed to agents with broad permissions destroying production data.
The fastest-growing category is also the one we’re most cautious about. OpenClaw, NVIDIA’s NemoClaw and Moonshot’s KimiClaw are persistent agent runtimes built around the same idea: always-on autonomous agents that can control computers. The security models do not hold up to the breadth of access these systems require, and we put all three in Hold. Where they’re already in use, sandboxing and human oversight are the only reasonable mitigations.
On the other hand, practices for agentic engineering are beginning to catch up. Spec-driven development enters Adopt in Techniques. In our experience, a structured specification, with guidance on trade-offs, architecture and goals, shapes and constrains an AI implementation more reliably than instructions written in prose. Formal specification languages bring a rigour often lacking from natural-language markdown specifications, and offer opportunities to give fast feedback on ideas and features before any code is written.
Two new Trial entries in Tools stand out, partly because we are seeing strong client demand for both: legacy modernisation and quality hardening. AI-assisted code migration takes the rote work out of porting codebases between languages, frameworks or major versions. AI red teaming tools probe deployed systems for prompt injection, data leakage and toxicity, adversarial testing that used to require a specialist consultancy on retainer. Both are cases of AI helping us produce more robust, better-tested software than most teams could justify writing by hand.
Ownership of models is moving back inside the organisation. The case for running open weight models on your own infrastructure has hardened: EU AI Act compliance requirements and the shift from flat-rate to usage-based AI billing both reward moving inference in-house. Small language models enter Trial in Platforms for the same reasons. Fine-tuned on a few thousand of an organisation’s own labelled examples, an SLM can handle classification and extraction work at a fraction of the API cost, on hardware already in place, with data that never leaves the network. For regulated or sensitive workloads, hosted APIs are no longer the only option.
Henry Garner (CTO, JUXT), June 2026
Radar Overview
Our radar is organised 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.
