Order healthy meals twice a day automatically
I've set up an agent to order food for me twice a day Every meal I've eaten over the last 2 weeks was decided by it, with the instruction to pick the healthiest options And I hate this thing so fucking much. Task: The user configured a custom agent to autonomously order meals twice daily for two weeks, strictly instructing it to select the healthiest available options. The agent handled the decision-making and ordering process without explicit human intervention for each meal, acting as a fully automated dietary manager. This workflow demonstrates an agent taking over routine personal consumption decisions based on a single health optimization goal.; tools: custom-built or unnamed agent. Human role: The original post does not say how a person stayed involved.; limitations: The source does not specify the underlying food delivery platform, API integrations, or payment methods used, creating uncertainty about technical feasibility and cost. There is no evidence of error handling for out-of-stock items or dietary restrictions beyond 'healthy,' suggesting potential gaps in robustness. Human approval steps were absent during the two-week trial, which may explain the negative user reaction to the lack of agency..
Tools
custom-built or unnamed agent
Published on Agent Work Atlas:
Overview
I've set up an agent to order food for me twice a day Every meal I've eaten over the last 2 weeks was decided by it, with the instruction to pick the healthiest options And I hate this thing so fucking much
Workflow
The user configured a custom agent to autonomously order meals twice daily for two weeks, strictly instructing it to select the healthiest available options. The agent handled the decision-making and ordering process without explicit human intervention for each meal, acting as a fully automated dietary manager. This workflow demonstrates an agent taking over routine personal consumption decisions based on a single health optimization goal.
Reported result
The user reported strong negative sentiment, explicitly stating they hated the experience despite the agent fulfilling its health-focused instructions. No quantitative metrics on meal quality, cost, or nutritional outcomes were provided, leaving the actual effectiveness of the 'healthiest' selection unverified. The primary outcome was user dissatisfaction with the lack of control or variety in the automated meal selection process.
Human role
The original post does not say how a person stayed involved.
Limitations
The source does not specify the underlying food delivery platform, API integrations, or payment methods used, creating uncertainty about technical feasibility and cost. There is no evidence of error handling for out-of-stock items or dietary restrictions beyond 'healthy,' suggesting potential gaps in robustness. Human approval steps were absent during the two-week trial, which may explain the negative user reaction to the lack of agency.
Try it
Act as a personal meal ordering assistant. I want to test an automated workflow for ordering lunch and dinner. My goal is to select the healthiest options from [replaceable: specific restaurant or delivery app, e.g., UberEats/DoorDash] within a budget of [replaceable: budget amount]. Steps: 1. Identify 3 healthy meal options for today's lunch and dinner based on my preferences: [replaceable: dietary restrictions, e.g., low-carb, vegetarian]. 2. Present these options to me for approval before placing any orders. 3. Once approved, simulate the ordering process and provide a summary of the selected items and estimated cost. 4. Do not actually place orders or spend money; this is a dry run. Output Format: - List of 3 lunch options with nutritional highlights. - List of 3 dinner options with nutritional highlights. - Awaiting my confirmation to proceed with the simulation. If any input is missing, ask me for clarification before proceeding. Ensure I remain in the loop for every decision to avoid the negative experience reported in similar cases.
Source reports are not independently verified. Publication date is the catalog publication date, not the original article date.