Creating a backlog of AI coding style tasks and then choose which to focus on right now.

Also some reflections on how to do AI coding atm.

πŸ’¬2025-12-27 have shiped Substack for Data and also did some work on reports that matter.

Summary

Choice between immediate intrinsic energy (vibe-coding on Reports That Matter) and longer-term strategic leverage (Substack for Data).

Consciously prioritising the latter: working on the Substack for Data plan because its importance and leverage.

Options

So I want to just organize a backlog of coding-related projects where I can try out my AI tools.

  • Life Itself website β€” web design and development.
  • Reports That Matter β€” publishing app for inquiries / public reports.
  • Flower Show: publish tool β€” CLI + workflow for rapid Markdown publishing.
  • Substack for Data β€” data-native publishing concept.
  • Florist β€” publishing app experiment on DataHub infra.

Strategic (separate)

  • Datopion: AI + data evolution β€” positioning Datopion around AI-and-data expertise.

Evaluation

Applied updates:

  • Renamed β€œForest” β†’ Florist
  • Life Itself website: Speed of feedback = 4, Scope size = M
  • Reports That Matter: Scope size = M, Speed of feedback = 3
  • Substack for Data: Speed of feedback = 5

AI Tool Experiment Backlog β€” updated scoring

ProjectType of practice (primary)Scope sizeLearning leverageSpeed of feedbackStrategic relevanceTotal
Life Itself websiteWeb design / web appM34512
Reports That MatterWeb app + publishing pipelineM4329
Flower Show: publish toolDeveloper tooling / CLIS55515
Substack for DataProduct + data publishingM45514
FloristData systems / publishingL43411

2025-12-22 - some further reflections

  • How can i have a sandboxed environment where i can 'yolo' AI code development
    • Remote options e.g. codespaces
    • Local options e.g. sandboxing via virtualserver

Re remote …

  • How do i use obra superpowers in codespaces
    • How do i automate setup of codespaces