AI Agent Use Cases
Views —

AI Agent Use Cases

Added today

44All 189

Prompt to Prepare a Shopping Cart from a Recipe Photo

The author reports that Grok Bot receives a recipe photo, organizes the ingredients and quantities, and creates a shopping list by excluding ingredients already available at home.

Grok Bot
Life & Hobbies

1–20 of 189

Marketing & DesignWalkthrough

Agent Team Creating Drafts for X, LinkedIn, and Instagram on a Single Topic

The author reports that a planning manager receives a topic and brand, organizes the message, and a writing manager creates posts for each channel.

CrewAI

How it works

  1. The author states that the topic to be covered is entered with --topic, and the brand and channels to write for are specified with --brand and --platforms.
  2. The planning manager organizes the core message, target audience, and an attention-grabbing introduction. The writing manager then receives these results and is asked to draft posts tailored to the length and format of each channel.
  3. The author notes that results are output to the terminal. There is no social account connection or posting function, so users review and use the text themselves as an example.

Results

This is an implementation example that generates channel-specific text based on a common plan. The author does not present any operational results such as increased views or engagement.

Notes

There is no code to fetch the body text from an entered URL. The character count condition for the original text is also just an instruction to the model, and the output length is not separately checked.

Source

GitHub Original post ↗

Example prompt · CrewAI

Set up a tool that creates post drafts for [channels to use] based on a [verified topic description] and [brand tone]. Refer to the planning and writing tasks in agents/14-social-media-agent, and first organize the Python environment, packages, and API settings.

In the planning stage, decide on the target audience and one core message. In the writing stage, adapt that message to each channel. Only use [figures, sources, and product descriptions] that I provide as facts, and do not assume the content of a URL if only the URL is given. Save the results for each channel in separate files, but do not automatically post to accounts.

Check that no performance results are fabricated from sources without numbers, that no unrequested channels are included, and that the specified character count for each channel is actually counted. The deliverables are planning notes, completed drafts for each channel, character counts, and a list of claims that need review. Distinguish any verification and file-saving features as additions in this implementation, not present in the original.