Sep 6, 20248 min read

Enterprise AI in Action: 7 Real Cases (feat. Danggeun Market, Dropbox, Geek News)

Is there really any company out there using AI well? We got curious. So we put together 7 real-world cases of how companies are actually using AI.

Uploading their own profile photo and asking for the review result.

This post is for you if
- You'd like to adopt AI but aren't sure exactly how to apply it
- You've wondered whether simple, repetitive manual work could be replaced by AI
- You wanted to see real examples of how AI is being applied at companies

?If we boil down the four things AI does well…
1. Summarization
- Geek News: AI summarizes newly posted Hacker News articles and uploads them to Geek News
- Nomad List: AI summarizes travelers' reviews and lays out the pros and cons on a single page

2. Category classification
- Danggeun Market (Karrot): AI reads people's titles and automatically classifies the product category

3. Review/proofreading
- Dropbox: AI proofreads every article according to the company's terminology rules
- Nomad List: AI reviews profile photos and blocks any that violate the guidelines

4. Chatbots
- Immi: An AI chatbot trained on the company's tone of voice handles 50% of CS inquiries
- Webflow: 65% of employees use an AI chatbot that searches internal documents

✏️ Let's walk through each AI use case one by one.

1. Summarization

AI is great at reading long pieces of information, pulling out the essentials, and summarizing them. It's perfect for processing data that's just piling up — data like customer reviews.

Case: Here's how other companies use AI summarization.

Geeknews

Service: A website offering development, tech, and startup news, visited by over 500,000 people each month.
Situation: When Geeknews staff or users found interesting tech news, they would write up the content and post it to the website.
How they use AI: They set up AI to summarize and translate some of the newly posted articles from sites like Hacker News so they could be published on the website. Today, more than 50% of the popular posts on Geeknews are AI-summarized articles. Thanks to AI summarization, the site can now publish far more articles, far faster.

On the left is the original article. On the right is the AI-summarized and translated version.

In September 2024, half of Geeknews's six most popular articles were AI-summarized.

Nomad List

Service: A community for nomad travelers, visited by 500,000 people each month.
Situation: Countless reviews left by people who had visited various cities kept piling up. Simply organizing those reviews well could become useful information in itself. But having a person summarize them every day was an inefficient task.

Reviews for each city kept piling up. There was a lot of useful information, but the posts were long and not easy to read.

How they use AI : AI summarizes the many city reviews so they can be read concisely. It became an important enough page to take up the third tab of a city's information.

An AI-summarized pros-and-cons comparison page for Bangkok

2. Category classification

One of the things AI does best is reading patterns. If you've had a task where a person had to assign categories one by one, there's a good chance AI can handle it well.

Case: Here's how other companies use AI classification.

Danggeun Market (Karrot)

Situation: Before AI was introduced, users had to choose the product category themselves. Recommending similar products wasn't easy, either.
How they use AI: They built AI that classifies product categories spot-on. When a product titled "New Balance 2002 RST 255" is posted, the AI defines the core category as New Balance. With categories classified, recommending similar products also became possible. Thanks to this, it's estimated that ad efficiency improved.

Danggeun Market uses the product-category information classified by AI in all kinds of ways.

3. Review and proofreading

There are plenty of review tasks that are minor but absolutely necessary — from deleting posts containing profanity to checking whether text is written according to the company's terminology rules. Review work with clear guidelines is one of the things AI does extremely well.

Case: Here's how other companies use AI review.

Dropbox

Service : A file-sharing service valued at 10 trillion won.
Situation : As the company grew, more and more members were writing content — from product FAQs to the tech blog, the business blog, and client stories. But because everyone used different terminology, communication became difficult. Consistency was needed. Yet asking members to just write according to the guidelines on their own was a tough ask.

FAQs, the tech blog, the company blog, client stories, and more — content was pouring in.

How they use AI : They built AI that automatically reviews and proofreads every sentence according to the company's terminology guidelines. They can now produce content far faster. Today, the 12 people who write content at Dropbox use the review AI to write.

They set up AI to automatically revise sentences to match the terminology guidelines.

Nomad List

Original situation: Because Nomad List is a community, keeping out shady people is extremely important. Until now, they hired part-timers to check each new member's profile one by one and take action if there were strange photos or behavior.
How they use AI: They used a GPT script to check newly registered profiles and flag suspicious ones. Thanks to the AI bot, they no longer need to hire part-timers.

They fed the review guidelines into the AI.

4. Chatbots

It seems we've reached a point where AI can write responses that no longer feel awkward. It's even more useful in industries with lots of repetitive questions. More and more companies are using AI to handle simple inquiries.

Case: Here's how other companies use AI review.

Immi

Service: A cup-ramen brand that has raised over 10 billion won. It's growing fast in the U.S. under the keyword of healthy, high-protein ramen.

Original situation: As the company grew rapidly, customer inquiries grew right along with it. More and more customers were waiting over 24 hours after leaving an inquiry. Rather than hiring more CS team members, they wanted to handle customer inquiries with a chatbot. Since the brand had a friendly image, it couldn't be a chatbot that answered in a stiff, robotic tone.
How they use AI: AI can now answer 50% of FAQ inquiries. Average response time got much faster, too. They moved to a setup where AI answers simple inquiries and CS team members handle the complex ones. What makes Immi's AI chatbot special is that it uses Immi's signature friendly tone. They trained the chatbot on that distinctive voice.

An AI chatbot explaining things just like a person.

Webflow

Service: A website-building service valued at over 5 trillion won.
Original situation: As the company grew rapidly, it became increasingly hard for employees to find the information they needed internally. It was difficult to find who built a product, in which department, and with what intent. It became hard to tell which document was the most recent, and which policies were no longer in use. There were more employees, but productivity dropped.
How they use AI: They built an AI chatbot that makes it easy to find the company's internal data. 65% of employees now use this chatbot. It's said to save about 300 hours a month that used to go into finding the documents people needed.

Asking the AI chatbot about information on a project run internally.

To sum up…

If there's a task a new graduate could perform without specialized knowledge, there's a good chance AI can do it too. In short, if your company has been hiring part-timers to do repetitive manual work, AI could likely help.

If you've been reading long pieces and summarizing them.
If you've been assigning product categories one by one.
If you've been doing repetitive review work to check whether things go out in line with specific guidelines.
If a person has been handling even the simplest CS inquiries.

Let's brainstorm AI adoption together with an expert!

Reach out to Dalpha, where AI experts come together. We'll consult on the AI that fits your company perfectly and build the solution for you. We'll get back to you within 10 minutes.

Apply for a free AI consultation

Bonus material

Here's the prompt Nomad List used to summarize reviews ?

The prompt Nomad List used for summarization

You are a moderator bot for Nomad List's city reviews. Nomad List is an unofficial travel community for digital nomads, and people leave reviews of cities. Here are the moderation guidelines.

1) Check for typical spammy reviews.
2) Disallow politics/racism/sexism/fighting.
3) Delete extremist takes such as the best place ever (fake) or the worst place ever (too extreme).
4) Profanity like Fuck, Shit, Shitty, Asshole, etc. is allowed. If it's a good review, there's no need to delete it.

If you approve the review, reply with 'YES'. If you don't approve it, reply with 'NO' and include the reason.