10 hours of Grok Bot usage exported: 89% went to background subtasks, 5 rules could cut 25-30%
95jun · reddit · 2026-10-22
A heavy user exported 10 hours of Grok Bot usage to CSV and had the bot analyze it, revealing where agent spend actually goes:
- 8,915 model calls, 1.15B tokens; direct chat was only 11% — the other 89% came from background subtasks the assistants spun up themselves.
- 96% of input tokens were re-reading old context, averaging 128K tokens of history per call; worst offenders were a 3D modeling task taking repeated screenshots and several PRs pushed in parallel.
Five rules he set for all assistants:
- Don't let a subtask wait over 5 minutes (cache expires and context gets rewritten, 13% of cost).
- Keep tasks small, wrap up around 120K context (calls over 150K were 54% of spend).
- Max 2 concurrent subtasks per assistant (peak was 23 at once).
- Poll every 5 minutes minimum, and only read log tails.
- Fewer screenshots at lower resolution.
Estimated savings: 25-30%, not yet measured. Takeaway: if you run agents, the cheapest win is exporting your usage data and actually reading it.
More from coding & agent
- Claude Code Pushes Urgent Security Update Across All Installations — umeshonai · 2026-10-23
- Claude cut a 2.6-hour video export to 20 minutes — but only after being told to rebuild the server — byhewar · 2026-10-23
- 3 flight-booking rules my AI agent checks that I'd never do by hand — Efistoffeles · 2026-10-23
- OpenJet claims Qwen3.8-27B coding runs 1.9x faster than llama.cpp on one RTX 3090 — JeremyCMorgan · 2026-10-23
- Building an autonomous SEO agent in Hermes with three skills: programmatic pages, Keywords Everywhere, DataForSEO — EXM7777 · 2026-10-23
- Building a sports betting AI agent with onchain execution and human approval by default: architecture and lessons — Dry-Comfortable-2678 · 2026-10-23