DEV Community

Cover image for Automate content and social media post idea generation - using Runner-H.
Fayaz
Fayaz Subscriber

Posted on

Automate content and social media post idea generation - using Runner-H.

This is a submission for the Runner H "AI Agent Prompting" Challenge

What I Built

We've all been there:

Completely stuck with writer's block, while trying to come up with social media posts for the day. 🥶

Whether you're building a personal brand or managing marketing content for your company or for your boss, fresh ideas from recent web articles can be a great source of inspiration.

But it gets even better with Runner-H.

In this post, I'll show you how to use Runner-H to search the web for new articles, collect them in a Google Sheet, summarize them in a Google Doc, and even generate sample tweets - all using a single prompt!

Demo

Runner-H Execution Demo

Runner-H Execution


Runner-H Web search Demo

Runner-H Web search Demo


Generated Google Sheet with article data

Generated Google Doc. with story by Runner-H

Generated Google Doc. with sample Tweets by Runner-H

How I Used Runner H

To use Runner-H properly like I've shown in this demo above, first you'll have to connect Runner-H with Gmail, Google Doc and Google Sheet.

This is not necessary to use Runner-H, but helpful if you want Runner-H to create those documents for you.

Once you log-in to your Runner-H account, go the the Connections tab and make necessary connections.

Runner-H Connections


Once your Apps are connected to Runner-H, go to the Run tab and use the Prompt below:

Objective: This task has 5 distinct but connected phases. The main goal is to:
- Collect recent posts from the specified online <source> on a specific <topic>.
- Save all the collected posts in a new Google Sheet.
- Write a report covering all the posts in a new Google Doc.
- Write social media posts in another new Google Doc based on the collected posts.
- Finally, email me the links to the generated Google Sheet and Google Docs.

Inputs: Some inputs will be provided at the end of the prompt within XML tags for clarity. Whenever an XML tag is mentioned, assume it will be provided at the end of the prompt.

Behavior:
- In each phase, you will behave like the world’s best expert on the subject matter.
- Try your best to accomplish the given tasks without user interaction.
- Plan the entire task before the actual execution that'll result in the best possible outcome.
- While executing the tasks as planned, if you find that executing the tasks as planned is not possible, amend the previous plan in a way that will satisfy the user’s main objective or at least come as close as possible within your capability.
- Prefix the name of each new Google Doc or Google Sheet created in this task with today's date. Date format: yyyy-mm-dd
- Format all the created Google Sheets and Google Docs beautifully, in a very professional and presentable manner.

Phase 1: Collect recent posts.
- Collect 10 recent posts on the given <topic> from the specified <source>.
- While gathering information on each post, collect the following data points: Title, Link, Published Date, Post Author Name, and Post Content.
- Don’t just collect information from search results; go to the original link of each post and collect actual post content from the main post.

Phase 2: Save all collected posts in a new Google Sheet.
- Once the necessary data of all the posts are collected, list them in a new Google Sheet with post Title, Link, Author, published Date, and Post Content.

Phase 3: Write a report in a Google Doc.
- Read the Google Sheet created in Phase 2, and based on the content of all the posts in the Google Sheet, write a report in a new Google Doc.
- This report should cover the main topics of all the posts.
- If helpful, link some of the posts within the content of the report.
- Write this report as a story, as if it's an observation and inner thoughts of a reader, while he is going through these posts starting a new day having his morning coffee. 

Phase 4: Write Twitter posts based on the collected posts
- Write 10 Tweets, 5 long form tweets and 5 short form tweets, based on the content of the collected posts.
- Don't write the Tweets like a copy or a report on the collected posts. Rather, write each Tweet uniquely, focusing on different ideas from the content of all those collected posts as a whole.
- Don't use any hashtag or any link within the Tweets.
- Give a separate heading to each Tweet in the Google Doc.

Phase-5: Email
- Once all above 4 phases are completed, email me the links to the created Google Sheets and Google Docs.

<topic>International Politics</topic>
<source>Google News Search</source>
Enter fullscreen mode Exit fullscreen mode

Note about the prompt:

  • The prompt is designed in a way that you'll only have to change the <topic> and <source> tags (presented at the bottom of the prompt) to fit it for your own needs!

  • Runner-H is in beta, so if one run doesn't generate the desired result, give it another try a few moment later.

  • Although I've asked it to use Google news search, it actually used Bing news search. May be it's related to its current preference. You can try with other web source.

Use Case & Impact

  • Anyone wanting to automate everyday tasks with Runner-H that's related to web search, story writing, social media post generation, writing Google Sheet and Google Doc. can get inspiration from my prompt.

  • Writing prompt this way and using it with Runner-H can help others create share-able automation techniques. This can really save a lot of time and help people spend more time on other important stuff, or even have more fun! 🥳

Social Love

Social Love 1: Sharing about Runner-H

Social Love 2: Sharing a sample Run of this prompt

Top comments (3)

Collapse
 
futuritous profile image
Futuritous

Interesting use of XML tags!
I'll definitely use this trick in other AI prompts.

Collapse
 
fm profile image
Fayaz

Thanks.
Structured prompts with XML tags work well in most of my use cases.

Collapse
 
fm profile image
Fayaz

More social love: