JUXT AI Radar (www.juxt.pro)

JUXT’s CTO Henry Garner has published the company’s first “AI Radar,” an opinionated, regularly updated guide that distils hands‑on lessons from JUXT’s year of building AI into client projects (coding assistants, agent frameworks, prompt engineering, model selection). The radar classifies technologies into four rings—Adopt, Trial, Assess, Hold—to indicate production readiness, and groups items across four categories: methodologies (how we design AI systems), programming languages/libraries/frameworks, development tools/utilities, and infrastructure/platform services. Each entry includes JUXT’s rationale for placement and practical advice, and the team invites community feedback via LinkedIn, BlueSky, and email. For AI/ML teams this is a practical heuristic for cutting through tool churn and marketing hype: it signals what JUXT believes is safe for production, what’s worth experimenting with, and what merits caution. Technically, the radar aids decisions on model selection, integration patterns (e.g., agent frameworks vs. assistants), preferred frameworks and dev tooling, and infrastructure choices—aligning architecture, governance, and risk management. Because it’s based on real engagements and will be updated, it’s useful both as a short‑term procurement/engineering checklist and as a living reference to track shifts in capability and maturity.