Getting started with GPTcsv

Welcome to GPTcsv! GPTcsv enables you to run AI — using any prompt — on your entire data set. You can use the power of GenAI to analyze, enrich, and generate data at scale.

Like chat-based AI products such as ChatGPT and Claude, GPTcsv is powerful and flexible — but it also takes some practice to get the best results. Let's walk through our first project together.

Example Project: What one thing would you change about Spotify?

GPTcsv has a free tier. You can create 1 project with up to 1,000 rows to get a feel for how it works. You can follow along as we walk through this example. After we're done, you can archive this project and create a new one using your own data or upgrade to create multiple projects.

We'll start with a dataset of 100 responses to the question: "If you could wave a magic wand, what one thing would you change about Spotify."

Now, we can create our project:

1. Go to https://www.gptcsv.ai/login to log into or create your account.
2. You'll receive a "magic link" in your email. Click on that link to login.
3. Click New Project in the upper right of your Project Dashboard.
4. Now, you can name your project and choose the .csv file you downloaded earlier.
5. Click Create Project when you are done.

After the .csv file uploads, you'll see a Project View that looks like this:

From here, you can browse the data in your .csv file using a familiar spreadsheet interface. You can sort and filter columns, and even create a Pivot view (we'll come back to that later).

Now, we're ready to work with the data. Instead of spending hours sifting through the survey responses, we can use Gen AI to label, evaluate, and prioritize the responses.

Your first Action Column

We'll use "Action Columns" to analyze, categorize, or generate data using AI. Every Action Column contains an AI prompt and can reference other columns in the same row.

Click Add Action Column to get started. You'll see the Add Action Column sheet:

From here, we can use a Saved Template to use an Action Column that was created previously or create a new Action Column:

1. Input a unique Column Title. Column Titles are used to reference the column in your prompt so pick a column title that is descriptive and unique.
2. Input a Prompt that will be used to generate this Action Column. We'll talk more about the Prompt below.
3. Optionally, click Save as Template if you'd like to save this Prompt to use in the future.
4. Select one or more Columns to include with the Prompt. These columns are the context for your Prompt. Select only the columns you need to reference in the prompt to get the best results. You can reference those columns by name in your prompt.
5. Select the Model you'd like to use for this Action Column. This is an advanced option and the default will work well for most cases.

In this case, let's create an Action Column that will classify each response into a topic that we can use to better understand the changes people want to see. This prompt will do a good job with that task:

You are a user research expert. I need help labelling user research responses for Spotify.

I will provide a user's answer to the question: "If you could wave a magic wand and change one thing about Spotify what would it be?"

Label the "What would you change about Spotify?" Column with a 1 word topic in all lower case for example, 'none', 'playlists', 'lyrics', 'performance', etc. Output 'none' if the column is 'none', 'not sure', 'idk', 'don't know'. '?', or any other unspecific response.

This prompt does a few things:

1. Establishes the role of the AI as an expert which — strangely enough — helps the AI generate better responses.
2. Provides useful context to help the AI understand what you are trying to accomplish with the task.
3. Mentions the specific column by name, "What would you change about Spotify?"
4. Gives clear criteria and some examples of good outputs.

Go ahead and create that Action Column now. Use the prompt above and make sure to include the "What would you change about Spotify?" column with the prompt. In a few seconds, you'll have 100 responses labeled by topic.

Analyzing the data

Now, we can analyze this unstructured survey data in a more structured way. For example, you can group the data by the newly created Topics column and see that most people would like to change things about the playlists on Spotify:

Refer to the video for additional details on how to use the analysis features in GPTcsv.

Editing the Action Column

You can rename, delete, or edit & re-run any Action Column. Click on the vertical dots — the "More" menu — to access those functions:

Evaluating response quality

Now that we have our Topics labeled, we can look at individual responses to get a sense of what types of things people would like changed about Spotify. Some of the responses are very useful, but others aren't great. Let's create an Action Column to rate each response by quality:

You are a user research expert. I need help rating the quality of a response to a user research question. I will provide a user's answer to the question: "If you could wave a magic wand and change one thing about Spotify what would it be?"

Rate the response to "What would you change about Spotify?" either "Good", "OK", or "Poor". A "Poor" response contains no useful information. An "OK" response mentions a specific feature. A "Good" response mentions a specific feature and provides an explanation.

For example, a "Good" response is:
"Better integration with Apple CarPlay. I would like to be able to more easily select songs from my playlist while I'm driving."

For example, a "Poor" response is: "Haven't used it enough yet to provide feedback" or "Nothing"

Respond only with "Good", "OK", or "Poor".

Similar to the prompt above, we use a few techniques to get a good response:

1. Establish the role of the AI as an expert
2. Provide useful context to help the AI understand what you are trying to accomplish with the task.
3. Mention the specific column by name, "What would you change about Spotify?"
4. Give clear criteria and clear examples for each of the valid outputs: "Good", "OK", and "Poor"
5. Give clear instructions to only respond with one of the three labels

Feel free to create the Quality Rating column now.

Prioritizing change requests

We've generated two columns, Prompt and Quality Rating. In a matter of seconds, we've enriched our survey data with information we can use to better understand what users want changed in the Spotify app. But, that's just the beginning.

We can tap into the full power of LLM's to work with our data — including the reasoning capabilities that are on the frontier of Gen AI. For example, we can use this prompt to prioritize change requests:

I am the product manager responsible for playlists at Spotify. My goal is to increase songs played per session. Evaluate each answer to "What would you change about Spotify?" based on the projected impact of songs played per session. Output "High" if the requested change will significantly increase songs played per session, "Low" if the impact will be none or minimal, and "Medium" if it will be somewhere in between.  

Go ahead and create this Action Column now. You may be surprised to find that the LLM does a good job figuring out which change requests will have the biggest impact on songs played per session. You can run this analysis as easily on thousands of survey responses or customer support messages — enabling you to quickly turn raw data into insights.

Explaining the prioritization

So, the downside of LLMs is that they can hallucinate — they'll provide an answer no matter what. Hallucination is even more problematic when we are asking the LLM to make a subjective selection from one of a handful of options. It'll pick one, and how are we to know whether that's the right choice or if the LLM just hallucinated something that sounded good?

Again, Action Columns to the rescue. Let's add a column to understand the reasoning behind each Priority recommendation:

I am the product manager responsible for playlists at Spotify. My goal is to increase songs played per session. Evaluate each answer to "What would you change about Spotify?" based on the projected impact of songs played per session. Output "High" if the requested change will significantly increase songs played per session, "Low" if the impact will be none or minimal, and "Medium" if it will be somewhere in between. I am the product manager responsible for playlists at Spotify. My goal is to increase songs played per session. You evaluated the answer to "What would you change about Spotify?" based on the projected impact to songs played per session and output a Priority of "High", "Medium", or "Low". Explain why the Priority is "High", "Medium", or "Low": what about the requested change will or will not have a significant impact on songs played per session? Output only the explanation, not the Priority. 

What's next?

So, there you go. In a few minutes, we've classified responses by topic, rated the quality of each response, assigned priorities based on the goal of increasing songs played per session, and generated explanations of those priorities.

Now, you can analyze the data using GPTcsv's built-in tools, export the data to bring it into your favorite data analysis tool, or use ChatGPT and Claude to generate a report — something that will turn out much better now that the data has been enriched.

We're here to help

Like other AI tools, GPTcsv is incredibly powerful, but takes some practice to get the most out of it. If you need some help, reach out at help@gptcsv.ai or book a call.

Example Project: Travel Destination Guides

In our first project, we used AI to analyze survey responses. We can tap into the power of Gen AI to not only understand language, but also to generate it. In this example, we'll start with a list of 100 of the most popular travel destinations.

Let's generate a travel guide for each destination listed in the file. We could simply add an Action Column where we ask the model to generate a travel guide. However, the results won't be great — they'll likely vary in quality and the content that's included. Instead, let's create some information that we can use as source material for the Travel Guide.

Generating travel categories

We can start by categorizing each destination by the type of travel popular in the destination:

You are an expert travel agent tasked with categorizing Destinations based on the type of travel common to those destinations such as 'Historical', 'Adventure', 'Relaxation', 'Shopping', 'Nature', 'Nightlife', etc. Output a comma-separated list of the travel categories that apply to the Destination.

This is a relatively simple prompt, but it does a few things that are important — it establishes the role of the LLM as an expert travel agent and provides some examples of good answers.

Go ahead, and try it now.

Generating "When to Go"

Now, let's start generating some content we can use as source material. We'd like to let people know when to travel to a destination:

You are an expert travel agent tasked with helping travelers determine the best time to visit Destination. Output an overview of the best time to visit per the example below.

Example:
Destination: Barcelona, Spain
Output: Summer (June to August) and fall (September to November): Summer is fiesta time in Barcelona, when the city hosts some of Europe's biggest music festivals, including Sonar and Primavera Sound. Average temperatures in summer have a high of 82°F (28°C) and a low of 71°F (22°C).

While soaring temperatures send summer visitors to the beach, the cooler months of fall are ideal for exploring Barcelona's colorful neighborhoods. In November, the scent of roasting chestnuts fills the air during the Catalan festival of La Castanyada. Average temperatures in fall have a high of 68°F (20°C) and a low of 60°F (16°C).

In this case, our example encourages the AI to be detailed and creative. Instead of just outputting the months, we help the AI be creative by including a well written example.

Generating more source material

Let's generate some additional content that is common to Destination Guides.

Things to Do

You are an expert travel agent tasked with listing the top Things to Do in a Destination. Output a comma-separated list of the top Things to Do that apply to the Destination.

Example:
Destination: Barcelona, Spain
Output: Basílica de la Sagrada Familia, Casa Batlló, Gothic Quarter (Barri Gotic), Park Güell, Mercat de la Boqueria, Palace of Catalan Music, Casa Milà - La Pedrera, Basílica de Santa Maria del Mar, Passeig de Gracia, Las Ramblas

Local Cuisine

You are an expert travel guide tasked with helping travelers experience the local cuisine in Destination. Write 1-2 paragraphs from the perspective of an expert travel guide advising a visitor.

Example:
Destination: Barcelona, Spain
Output: Barcelona is renowned for its vibrant and diverse culinary scene, deeply rooted in Catalan traditions while also reflecting Mediterranean influences. One of the must-try dishes is pa amb tomàquet, a simple yet flavorful starter made of toasted bread rubbed with ripe tomatoes, olive oil, and salt--often served with cured meats or cheeses. Another essential dish is escudella i carn d'olla, a hearty Catalan stew that is especially popular in winter. It's made with a variety of meats, vegetables, and pasta, offering a true taste of home-cooked tradition.

For seafood lovers, Barcelona's coastal location ensures fresh catches daily. Suquet de peix, a Catalan fish stew, combines fish, potatoes, and saffron in a rich broth. The city is also famous for calçots, a type of green onion grilled over an open flame, typically eaten with romesco sauce during the winter calçotada festival. To experience Barcelona's bustling food culture, a visit to La Boqueria Market is essential. There, you can sample an array of local delicacies and soak in the vibrant atmosphere of one of the city's most iconic culinary spots.

Generate the Travel Guide for each destination

Lastly, we can use all of the source material to generate a rich, detailed, creative travel guide for each of the 100 destinations. Our prompt doesn't need to be too detailed because we have some much source material to work with:

You are an expert travel guide tasked with helping travelers plan a great trip to Destination. Write a 3-5 paragraph guide to the Destination using all of the information provided including Things to Do, Best Time to Travel, and Local Cuisine. 

Here is a couple examples of Travel Guides that were generated:

Rome, Italy

Welcome to Rome, Italy, a city where ancient history meets modern vibrancy, offering travelers a rich tapestry of experiences. As you explore the historical wonders of Rome, you'll find yourself surrounded by monumental landmarks that tell the stories of eras long past. The Colosseum stands as a testament to ancient Rome's architectural prowess, inviting visitors to step back in time and imagine the gladiatorial contests once held within its walls. In the Vatican City, the Vatican Museums await, home to one of the most impressive art collections in the world, culminating in the awe-inspiring Sistine Chapel, where Michelangelo's masterful ceiling frescoes can be admired. The Pantheon, with its iconic dome and oculus, and the Roman Forum's archaeological marvels also invite exploration, making Rome an unforgettable journey through history.

When planning your visit, consider the seasonal charm of Rome. Summer, from June to August, is bustling with outdoor events that bring the city streets to life, yet be prepared for the warm temperatures, often reaching 86°F (30°C). Alternatively, spring, spanning from March to May, offers a more temperate climate with highs around 71°F (22°C). This season also brings Rome's gardens into full bloom and offers a plethora of cultural events, such as the lively Rome Spring Festival, providing the perfect opportunity to experience Rome's vibrant arts and community scene.

No trip to Rome would be complete without diving into its culinary delights. Known as a culinary paradise, Rome offers a variety of traditional delicacies that are sure to tantalize your palate. A must-try is cacio e pepe, a simple yet flavorful pasta dish featuring Pecorino Romano cheese and black pepper, showcasing the art of Roman cooking. Supplì, a scrumptious fried rice ball filled with mozzarella, epitomizes the essence of Roman street food. To truly immerse yourself in Roman culinary culture, visit a classic trattoria for saltimbocca alla Romana, featuring tender veal, prosciutto, and sage harmonized with white wine. Lastly, explore Rome's lively open-air markets, like Campo de' Fiori, where fresh ingredients tell the story of Roman cuisine. End your culinary journey with a serving of creamy gelato, a sweet Italian ice cream, perfect for rounding off your Roman feast. With its incredible history, cultural events, and mouthwatering cuisine, Rome promises an enriching travel experience you'll long cherish.

Bangkok, Thailand

Embark on an unforgettable journey to Bangkok, Thailand, a city that artfully merges age-old traditions with modern dynamism. Known for its vibrant culture, Bangkok offers an array of experiences that cater to diverse travel interests. Dive into the local life by exploring bustling shopping hubs and experiencing the city's renowned nightlife. Bangkok is also a hub for cultural enthusiasts, with myriad historical sites and temples to explore, such as the iconic Grand Palace and the enigmatic Wat Pho. Each visit tells a story of Thailand's rich heritage, while spots like the Chatuchak Weekend Market and Khao San Road provide a glimpse into the city's eclectic mix of modernity and tradition.

Timing is everything when visiting Bangkok. The best period to explore this magnificent city is during the cool, dry season between November and February, when temperatures are more manageable. With highs around 85°F (29°C) and lows of 70°F (21°C), travelers can comfortably explore Bangkok's streets and landmarks. This season coincides with vibrant cultural festivals like Loy Krathong, which transforms the city into a luminous spectacle of floating lanterns and festive activities, providing unique cultural experiences and opportunities to delve deeper into Thai traditions.

No trip to Bangkok would be complete without indulging in its world-renowned street food. The city is a haven for food lovers, with tantalizing flavors bursting from every corner. Start your culinary adventure with pad thai, an iconic noodle dish that balances sweet, sour, and savory notes. Then, dive into the zest of som tam, a fiery green papaya salad that perfectly harmonizes tangy lime and spicy chilies. For those craving something warm, tom yum goong offers a delightful spicy shrimp soup, while mango sticky rice serves as a refreshing dessert, coupling ripe mango with coconut-drizzled sticky rice. To experience Bangkok's culinary diversity firsthand, venture into the ever-busy Chatuchak Weekend Market or the vibrant Yaowarat Road, where a plethora of vendors showcase Bangkok's rich culinary heritage.

Try it for yourself

In a few moments, we've generated a 100 pages of rich travel guide content. You can continue to work with the travel data set or you can generate your own content. We actually used ChatGPT to generate the .csv file we used for this project using a simple prompt:

List out the 100 top travel destinations. Output a .csv file.

You could do the same for songs, books, movies, restaurants, or any other collection. From there, you can use GPTcsv to generate content at scale — using whatever prompts you can imagine.