Grok BotUndisclosed

Uber driver uses Grok Bot to generate a daily one-page drive plan and take/avoid rules from historical offer data

Levi SpiresX (Twitter) · 2026-11-05

Read original

Uber has an algorithm. So do I. Here’s how I use Grok Bot as an Uber driver. Before I wake, my Driver Grok Bot builds me a one-page drive plan from my own data: 12,291 Uber offers, heat maps of where to focus and avoid, GigU app acceptance floors, that day’s flights, road work, and local events. Then it turns that into take/avoid rules and where to stage. It took about a day to vibe code this. And takes about 30 seconds to use it. Why isn’t every business operating this way today?

Business model

Independent Uber driver Levi Spires has a custom Driver Grok Bot that, before he wakes, builds a one-page drive plan from his own 12,291 Uber offers, heat maps, GigU acceptance floors, that day’s flights, road work, and local events. It converts this into take/avoid rules and staging locations. Setup took about a day; daily use about 30 seconds. Readers can reuse the pattern of feeding personal ops data to an Agent for daily executable rules.

Workflow

  1. Import personal historical Uber offers data, heat maps, and GigU acceptance floors into Grok Bot context
  2. Aggregate daily external signals (flights, road work, local events) and generate a one-page drive plan
  3. Convert the plan into specific take/avoid rules and staging locations

Adaptation plan

Results and time claims are as reported by the original post author. Workflow details, freePrompt, and trial I/O are AgentFlow reconstructions and were not live-tested here. No income claimed. Grok Bot subscription amount undisclosed; exact data access method and vibe-code details not shown. Some $/hour figures at the bottom of the first attached image were partly obscured and were not used.

Prompt

AgentFlow draft · Untested.

You are my Driver Ops Agent. Generate today’s one-page drive plan from the materials I provide:
- Historical offer data summary (at least date, zone, duration, earnings or offer count)
- My acceptance floor (e.g. minimum hourly or accept rate)
- Today’s external signals (flights, road work, events list; mark missing if none)

Task: 1) Summarize high- and low-value zones; 2) Give 3-5 take/avoid rules; 3) Suggest 2 staging locations.
Output format: Markdown one-pager with zone table, rules list, staging suggestions.
If data is missing or conflicting, list the gaps explicitly and give only conservative advice based on available info. Do not invent numbers.
Before suggesting any send, booking, or external action, list it and wait for my confirmation.
This is an untested AgentFlow reconstruction prompt.

Running costs and pitfalls

Results and time claims are as reported by the original post author. Workflow details, freePrompt, and trial I/O are AgentFlow reconstructions and were not live-tested here. No income claimed. Grok Bot subscription amount undisclosed; exact data access method and vibe-code details not shown. Some $/hour figures at the bottom of the first attached image were partly obscured and were not used. No independent cost benchmark is available. Check subscriptions, permissions and external actions before running a workflow.

Author-reported results are not independently audited.

Original text and media remain attributed to their authors.