Book weekly restaurants and surface daily discoveries from your data
I gave OpenClaw access to all my ChatGPT, Google and Instagram history via context-use: - It books weekly restaurants for me using my years of Instagram stories - last night I went to Osteria Vibrato which was discovered and booked by my agent - Sends me 3 interesting discoveries every day based on what I am doing on GPT/Instagram/Google search. Task: The author connected OpenClaw to ChatGPT, Google, and Instagram histories using context-use. The agent analyzes Instagram stories to book weekly restaurants and curates daily discoveries based on search activity. The user manually reviews and approves daily memory diffs to maintain data accuracy.; tools: OpenClaw. Human role: The original post does not say how a person stayed involved.; limitations: The author is part of the context-use team, introducing potential bias. The post does not specify costs, privacy safeguards, or error handling for bookings. Reliance on historical data may limit novelty. Human approval is required for memory updates..
Tools
Published on Agent Work Atlas:
Overview
I gave OpenClaw access to all my ChatGPT, Google and Instagram history via context-use: - It books weekly restaurants for me using my years of Instagram stories - last night I went to Osteria Vibrato which was discovered and booked by my agent - Sends me 3 interesting discoveries every day based on what I am doing on GPT/Instagram/Google search.
Workflow
The author connected OpenClaw to ChatGPT, Google, and Instagram histories using context-use. The agent analyzes Instagram stories to book weekly restaurants and curates daily discoveries based on search activity. The user manually reviews and approves daily memory diffs to maintain data accuracy.
Reported result
The author reported visiting Osteria Vibrato, which was discovered and booked by the agent, and enjoying a book recommendation for a Kerala trip. No quantitative metrics or failure rates were provided. Actual performance consistency remains unknown.
Human role
The original post does not say how a person stayed involved.
Limitations
The author is part of the context-use team, introducing potential bias. The post does not specify costs, privacy safeguards, or error handling for bookings. Reliance on historical data may limit novelty. Human approval is required for memory updates.
Try it
Act as my personal assistant using OpenClaw. I will provide access to my [ChatGPT/Google/Instagram] history via context-use. First, analyze my last [30 days] of Instagram stories to identify dining preferences. Propose one restaurant booking for [next Friday] and three daily discoveries based on my recent searches. Present these to me for approval before executing any actions. Do not send messages or confirm bookings without my explicit 'yes'. If data is missing, ask for specific date ranges. Show me the proposed memory diff for review before saving. Confirm you understand the human-in-the-loop requirement before proceeding.
Source reports are not independently verified. Publication date is the catalog publication date, not the original article date.