I cut my AI coding bill and doomscrolling by running one nightly shutdown ritual

Source: Dev.to

Problem

• My AI coding spend kept creeping up every week.
• My nights disappeared into feeds, tabs, and “just one more scroll.”

I thought these were two separate issues, but they were symptoms of a single system failure. When my attention was fragmented, my prompts suffered, leading to higher costs and lower productivity.

Root Causes

• Re‑running prompts with fuzzy scope.
• Constant context switching between coding and endless feeds.
• Using heavyweight models for lightweight tasks.

Nightly Shutdown Ritual

1. Review session costs – Identify which sessions burned the most tokens and write a brief note: “Why was this expensive?”
2. Pre‑decide model usage – Determine in advance which tasks deserve premium reasoning and which can stay on cheaper/default models.
3. Eliminate algorithmic feed loops – Instead of deleting apps, I remove the feed loops that hijack context, allowing me to start with clear prompts rather than morning chaos.

Tools I Built

• TokenBar ($5, one‑time) – Shows live token and cost visibility in the macOS menu bar while I work.
• Monk Mode ($15, one‑time) – Blocks feed‑level distractions, preserving utility while eliminating the infinite‑scroll trap.

Both were created for personal use and have become the backbone of my nightly routine.

Results

• Fewer “mystery” cost spikes.
• Less retry churn on prompts.
• Cleaner deep‑work sessions at night.
• More shipped work with less mental drag.

Takeaway

If your AI spend feels random, examine your focus system before blaming model pricing. A disciplined shutdown ritual can dramatically reduce costs and improve productivity.