Using AI to handle your most annoying data tasks
An AI data assistant can be a reliable (and fast) way to remove small data annoyances that crop up during any given task.
Repetitive data tasks are the bane of many an existence. The worst thing about them? They vacuum up time without you even noticing. Maybe you’ve tried creating a meeting for fifty folks, and have had to add emails one-by-one. Or perhaps you’ve wanted to take a list of ideas and put them into a spreadsheet, and even after pasting you’re left cleaning up each cell.
An AI data assistant can be just the way to keep yourself free and clear from that inglorious muck. Below, we’ll look at a series of prompts to do just that, as well as a flexible prompt that can be reused, regardless of your inputs and outputs.
Key Takeaways
AI can squash the time it takes to do data tasks that crop up on the fly
Think of it less as a highly-skilled assistant, and more as a container for these annoying tasks
One flexible prompt can help you convert any data (by passing the data and your input and output formats)
But first, a quick qualification of what I mean by ‘data tasks’. I’m referring to those small data annoyances. The ones that crop up between you and your actual goal. They’re ugly but necessary but ever-present, and important for the real work while simultaneously getting in the way of it.
You know of whom I speak. They can be anything from:
Turning a bulleted list into a table (so it’s easier to paste into a spreadsheet)
Taking a long email that’s kind of organized and converting it into sensible rows
Organizing unformatted ideas (articles, comments, feedback) into a homogenous format
Converting a bunch of headlines from title case to sentence case
Getting a list of emails from the “To:” field (while sifting out all that other crap)
Taking a few hundred rows of data and converting it to JSON format
This is a short subset of a very long list, all of which AI can help with. To be clear, what I am NOT referring to is advanced excel formulas, ETL processes, or anything more data heavy. If that’s why you’re here, well… hopefully this still helps.
Why use an AI data assistant?
Think of your “assistant” less as a highly skilled craftsman and more as a container for these tedious tasks. A single place you can take ugly data and get it back prettier without the extra effort. I, for one, dedicate a single ChatGPT thread to all of these tasks.
Ultimately, using AI for small data tasks isn’t about doing things you couldn’t do manually. It’s about removing the manual work and retaining the uninterrupted flow of your actual work. Some other benefits include:
Reducing repetitive work so you can focus on the strategic
Increasing accuracy of data conversion without sacrificing your brainpower
Simple data tasks
I’m going to start by overwhelming you with a number of examples I’ve used previously, many of which you can copy for your own specific use. Below you’ll find prompts to:
Put a bulleted list into table format
Turn unformatted content into a bulleted list
Turn bulleted list into table
Combine tables into a single table
Pull a list of email addresses from the “To:” field
Pull a list of URLs from article
Turn a markdown post into HTML
Take a list and return a two-column table, organized by column
Pull a list of fields from web page content
Put unstructured (or table) data into JSON format
Data conversions, especially JSON transformations, can be repetitive. Let’s say you’re working with player statistics.
Please the take the following statistics (for a single game) and convert it into JSON.
-- stats START --
Deshaun Watson #4 Dorian Thompson-Robinson #17 Jameis Winston #5 C/ATT YDS AVG TD INT SACKS QBR RTG 15/17 128 7.5 0 0 2-13 41.1 98.0 11/24 82 3.4 0 2 1-3 15.4 19.8 5/11 67 6.1 1 0 1-2 75.0 95.6
-- stats END --Ultimately, using AI for small data tasks isn’t about doing things you couldn’t do manually. It’s about removing the manual work and retaining the uninterrupted flow of your actual work.
Put a bulleted list into a table format
This prompt allows for the transformation of unstructured lists into a structured, alphabetized table.
Please take the following list and provide this list back to me as one long, uncategorized table with the column header "{COLUMN_HEADER}". Additionally:
* For any content in parentheses (example: "(aka Twenty/Twenty Literary Group)", please add that to a column B for that row.
* List all rows in alphabetical order by {COLUMN_HEADER}
COLUMN_HEADER = "[ex. Name]"
-- List START --
#
– 3+ Creative Media
– 20/Twenty Literary Group (aka Twenty/Twenty Literary Group) (see Bridgebooks)
A
– A-Plus Task Masters
– Access Media Group (aka Quill Space Media)
B
– Bantam Wings (see Get Started Books)
– Beacon Books Agency (aka Fact & Fiction Entertainment and Literary Agency)
—- List END —-Turn unformatted content into a bulleted list
This prompt converts lengthy text into an easy-to-read bulleted format. In the case below (used for a glossary), each item follows a pattern ("{TERM}: {Definition}").
Please take the following content and turn it into a bulleted list. When formatting my list:
1. Each bulleted must be formatted as "{TERM}: {Definition}"
2. All bullets must end with a period
-- Content START --
Pun
This is a play on words with a similar spelling or pronunciation, but different meanings. A pun exploits both meanings. Here’s an anonymous example: Seven days without food makes one weak.
Alliteration
This stringing together of words that start with the same letter is orally pleasing and quirkily humorous. Write with whatever words will work!
Acronym
Make one up, or redefine one. LOL might mean Little Old Leprechauns.
Anachronism
Thou must have no chill with this, but make sure you do it on purpose. In Caesar, Shakespeare mentions a clock, which had not yet been invented.
Malapropism
Using words incorrectly is perfectly demonstrated by Cormac McCarthy in his dark novel Child of God, where an illiterate dumpkeeper names his daughters based on an old medical dictionary. If you’re having a baby, consider these names from McCarthy: Urethra, Cerebella, and Hernia Sue.
-- Content END --Turn a bulleted list into table format
Convert each list item into table rows with multiple columns. This provides a structured view of text-based data, providing consistency.
Please take the bulleted list of data below and turn it into a tabular format. * I want you to take that data and give me a table with the columns: {COLUMNS}. Additionally:
* I'm giving you a list of literary agents, each of whom is at a different agency
* The notes column will include all of the data for each record that is NOT the agency or agent name.
COLUMNS = “[‘Agency’, ‘Agent Name’, ‘Notes’]”
-- Bulleted list START --
The Cheney Agency
Elyse Cheney
Location: New York, US
Genres: Historical Fiction, Literary Fiction, Commercial Fiction, Nonfiction, Memoir, and more.
Notable Clients: Dave Eggers, Terri Jentz, Benjamin Kunkel, Nathaniel Rich, Suzanne Rivecca.
Website: Cheney Agency1
HG Literary
Brianne Johnson
Location: New York City, US
Genres: Historical Fiction, Women’s Fiction, Young Adult, Children’s Book, Diverse Fiction, LGBTQ, and more.
Notable Clients: Wesley King, Fran Manushkin, Margo Kelly, Andrew Root.
Website: HG Literary2
YRG Partners
Yfat Reiss Gendell
Location: New York, US
Genres: Historical Fiction, Science Fiction, Thriller, Women’s Fiction, Nonfiction, Memoir, and more.
Notable Clients: Ernest Cline, Denise Kiernan, Karla Cornejo Villavicencio, Reese Witherspoon.
Website: YRG Partners3
-- Bulleted list END -- Pull a list of email addresses from the “To:” field
When creating an event or starting a new email thread, it can be way to hard to copy email addresses (if your email client is good with this, I demand you tell me now). Copy the mess of names and emails, and AI can return only the addresses.
Please pull only the list of emails out from the following data so that I copy and paste them into a new email.
-- Emails END --
"John Doe" <j.doe@company.com>,
"Jane Smith" <jane.smith@example.com>,
"Sam Rivers" <s.rivers@company.com>,
"Lily Jones" <lily.jones@example.com>,
"Chris Perry" <c.perry@company.com>,
"Marcus A. Brown" <marcus.a.brown@gmail.com>,
"Nina Gomez" <n.gomez@company.com>,
"Oliver Williams" <oliver.williams@example.com>,
"Paula Wright" <p.wright@company.com>,
"Emma Clark" <emma.clark@workmail.com>,
"Tyler Shaw" <t.shaw@company.com>,
"Molly White" <molly.white@example.com>,
"George King" <g.king@company.com>,
-- Emails END --Combine tables into a single table
Merges two tables with similar structures into one, allowing for consolidated data. It’s at its best when you have two different but similar tables.
Please combine these two tables into a single table? In other words, I want a single table that includes both the first list of novels released January 2024, and the second list of novels released February 2024. The structure of the table and all of its columns MUST remain the same.
-- Table 1 START --
[Copy the content from Table 1]
—- Table 1 END --
-- Table 2 START --
[Copy the content from Table 2, best when in same/similar format to Table 1]
-- Table 2 END --Pull a list of URLs from article
Extracts company names and URLs from a block of text, creating a straightforward list of resources, which is helpful for quick access or further exploration.
Please parse through the below article to provide me only the names and the URLs of included in the article. You must provide it back to me as a list of text broken by line breaks.
-- Article START --
[These companies aren’t just growing fast locally and garnering…]
-- Article END -- Turn a markdown post into HTML
If you ever have markdown-formatted content (which isn’t friendly to copy/paste) and want to paste it somewhere else, an HTML format is much more friendly. Very helpful when moving content from one system to another.
Please take the below blog that is in markdown, and repaste it back to me so that I can copy it and directly paste it into a substack draft.
-- Blog post START --
In an age where artificial intelligence is increasingly integrated into our daily lives, it's crucial to understand the layers of security that protect our data, especially when interacting with AI systems.
### Layers of Security in AI Systems
When we talk about AI security, we're really talking about how safe your data is when you're using AI tools. Think of it like a series of locked doors that only you have the keys to, and these doors protect everything you share with the AI.
**Systemwide Security**: This is like the big, sturdy walls around your home. It’s what keeps your data safe from outsiders.
### What I Will Not Share with AI
On the flip side, there are clear boundaries when it comes to what I won’t share with AI. Here’s a list of what I will not share:
1. Unpublished manuscripts.
The key is to stay informed and proactive, ensuring that our use of AI remains both productive and secure.
-- Blog post END --Take a list and return a two-column table, organized by column
Sorts and organizes a list by name in ascending order, with an added feature to assign teams to specific columns, facilitating quick searches and a structured layout for easy reading.
Please take the following list of NFL teams grouped by BYE WEEK and provide me back a table of this data, organized by team name ASCENDING. The team "Arizona Cardinals" should be first and "Washington Commanders" should be last.
-- NFL teams by Bye Week START --
Week 5
Detroit Lions
Los Angeles Chargers
Philadelphia Eagles
Tennessee Titans
-- NFL teams by Bye Week END --Pull a list of fields from web page content
Have a form or settings page to fill out? You can extracts fillable fields and from a web page. As a bonus, you can get help in generating the content for these fields by adding to this prompt (“I will also want you to generate content for each of these fields.”)
I have copied an entire web page of settings that I need to fill out. I need you to parse through everything I've copied and pasted to pull out:
1. All of the fields I will need to fill out and
2. All the checkboxes I need to select.
-- Web page content START --
Publication name
AI Aristry
Short description
My personal Substack
This appears at the top of the welcome page, and should succinctly explain what you write about. New visitors are automatically shown this page if they haven't been to your publication before.
Site Design
Set color, layout, and font styles for your site.
Customize
Language
This is the language for your posts, website, and all reader communications
English
About page
This page should explain in detail the benefits of reading your publication.
Edit
Square logo
-- Web page content END --A data conversion power prompt
All of the above examples work around a similar concept. I'm going to provide three things:
the data to handle
the current format of your data
your desired output format
Given that, here’s an incredibly flexible prompt that can accomplish 90% of your menial data needs, whether you’re handling JSON, CSV, or XML.
Convert the following data from {INPUT_FORMAT} to {OUTPUT_FORMAT}. Please ensure maximum accuracy in this conversion.
INPUT_FORMAT = "[Table/Bulleted List/Paragraph/JSON/XML/etc.]"
OUTPUT_FORMAT = "[Table/Bulleted List/Paragraph/JSON/XML/etc.]"
-- Data START --
[Paste your data here]
-- Data END --