Technology RadarTechnology Radar

AI Agentic Coding Tools

ideaiagenticproductivity
Assess

Reason for selection

Besides the obvious hype, agent-assisted IDEs promise major speed and quality gains. Picking the best-in-class tools and standardising workflows should keep MOHARA competitive and accelerate delivery.

Key questions

  • Can we build complex apps end-to-end in an agentic flow?
  • Which IDE offers the best developer experience (DX)?
  • Does AI effectiveness drop as a project matures?
  • What "works" when coding with agents—best practices, common pitfalls?

Assessment approach

  1. Desk research on current best practices.
  2. Build the same small project in Cursor, Windsurf, Zed, VS Code + Copilot, etc.
  3. Optionally pilot on real projects and capture metrics.
  4. Consolidate findings into a best-practice guide.

Potential testers

  • Any engineer keen to try AI IDEs.

Notes

Could be run as an open bake-off if we define a tight rubric.