Grok Bot Savings Tracker
I spend $300/mo on Grok Bot.
I'm tracking how much money it saves me every month: every real task my bots finish, priced at what a freelancer would charge.
Month 1 · Nov 2 to Dec 1, 2026
Day 1 of 30
Month 1 is in progress. The real scorecard is the full month against the full $300, posted after Dec 1. Data as of Nov 2, 2026.
Spend
$300
$200 + $100 plans · $10.00 prorated to day 1
Savings est.
$1,009.50
29 h of freelancer time · 17 credited tasks
Revenue earned
$0
Received money only. Nothing yet.
Total return
+$709.50
savings + revenue − $300
Savings are 3.4x the full $300 month (101x vs prorated spend). Against spend prorated to day 1, the total return is +$999.50.
By role
Who I didn't have to hire
What each role would cost for the same hours, at the rates below.
By business
Where the work went
Task log
Every task, priced
8am briefing: GSC pull + rundown, X performance
$25
Daily post drafts
$0
Competitor URL snapshot (bimmerbench)
$17.50
Mitsubishi decoder: OEM spec-sheet + warranty research
$40
Mitsubishi decoder page: build, deploy, rollout notes (PR #20)
$40
Thin-page diagnostic crawl
$30
Noindex pages.dev preview copies (PR #35)
$35
X5 air suspension compressor brief + ideas board update
$35
X5 compressor sections: owner prices + G05 parts listings
$40
3 text-screenshot ad scripts
$70
Rendered ad images (v1 + v2 stories)
$50
Togi $10M Challenge tracker: Worker + /togi/ page (PRs #14-16)
$160
v3: pen-name check, copyright page, rebuild
$50
v4 digital edition: linked 89-page PDF + 40-page worksheets pack
$250
Sales platform research
$57
Evening digest with fact-check
$25
7 reply + quote-tweet drafts (v2 credited)
$25
Built this tracker: rates research, ledger, Page
$60
Misses and redos
What went wrong
Quote targets that were already quotes
RedoSome of tonight's v1 quote-tweet picks were already quote tweets (Garry, Tony, levelsio). The v2 set still includes a Jason post that quotes @chamath.
Weak YC pick
LearningThe YC post was a weak quote target. Maxwell had to ask why it was suggested.
Quotes too long
RedoThe v1 quotes ran long. Quotes need to be short and punchy, and replies can be longer.
Quote-tweet composer link
Process fixThe composer link opens a normal post with the tweet URL in it instead of a real quote tweet.
Morning drafts unused
$0 creditedToday's 8am drafts weren't posted. They're logged at $0 until they're used.
Tracker rebuilt
RedoThe first version backfilled Oct 15 to Nov 1. It was rebuilt so tracking starts today.
Redone work is credited once. Unused work is logged at $0.
Methodology
How the numbers work
Value is hours times rate. Hours are a conservative guess at what a competent freelancer would bill, not how long the bot took. A task is logged only if a merged PR, a file or a bot log entry backs it. Code written by Cursor cloud agents, or built and deployed by the bots themselves, gets full developer credit. Code written by Claude Code or Codex gets only spec-and-review time at the project manager rate. Unused drafts and abandoned work are logged at $0 until they're used.
Revenue counts only money actually received, with evidence (a payment, payout or royalty record). Pending, invoiced, pledged or projected money never counts. Savings and revenue are kept as separate figures, and total return is savings plus revenue minus the $300 spend.
Month 1 runs Nov 2 to Dec 1, 2026 (30 days). Spend is $300 for the month: $200 + $100 plans. Nothing before Nov 2 counts. Rows marked “inferred” credit the bot that most likely did the work. The plan's usage is shared with other apps, so some of the $300 may pay for things this tracker doesn't cover. Updated weekly.
Mid-range US freelance or median rates, loaded Nov 2, 2026. Sources: Upwork: Web developer rates · Upwork: SEO expert rates · Upwork: Content writer rates · Upwork: Social media manager rates · Upwork: Virtual assistant rates · Upwork: Data analyst rates · Upwork: Graphic designer rates · Editorial Freelancers Association: Editorial rates chart · BLS: Project management specialists pay · BLS: Market research analysts pay.