Three-agent real-world test: only Grok Bot finished the complex travel concierge job
Shane KidwellFacebook · 2026-10-16
Read originalSaved text / editorial summary
I've been running three AI assistants side by side... Muse... ChatGPT Work... Grok Bot... Then I ran a test. I set up a travel concierge and asked all three to scrub every travel site plus my Amex travel account. Only Grok Bot finished the job.
Business model
Operator ran Muse, ChatGPT Work and Grok Bot side-by-side during a Hawaii trip. Grok Bot scrubbed 5 email accounts and managed social engagement. In a travel-concierge test requiring scrubbing every travel site plus an Amex account, only Grok Bot completed the full job.
Workflow
- Three agents run in parallel on daily executive admin and social tasks
- All three agents are asked to scrub every travel site + Amex travel account in a travel concierge test
- Author documents real performance differences as a decision basis
Adaptation plan
Results are as reported by the original post author. Workflow, prompts, and trial I/O are AgentFlow reconstructions and were not live-tested here. No income claimed. Exact success criteria and intermediate failures of the other two agents not fully detailed; no long-term cost comparison. Post date per original page 2026-10-16 (batch stated 2026-10-17).
Prompt
AgentFlow draft · Untested.
Please run a side-by-side comparison of multiple agents. Give them a complex task that requires crossing multiple websites and accounts (e.g., scrub travel-related accounts and produce a usable plan). Record how far each agent gets and rank reliability. Untested reconstruction based on the original post.
Running costs and pitfalls
Results are as reported by the original post author. Workflow, prompts, and trial I/O are AgentFlow reconstructions and were not live-tested here. No income claimed. Exact success criteria and intermediate failures of the other two agents not fully detailed; no long-term cost comparison. Post date per original page 2026-10-16 (batch stated 2026-10-17). No independent cost benchmark is available. Check subscriptions, permissions and external actions before running a workflow.