design

How We're Using Generative AI To Improve User Experience

How We're Using Generative AI To Improve User Experience

Give it a read if you want to know how we're using small generative AI features to bring a big impact to the overall user experience of Peerlist.

Akash Bhadange

Akash Bhadange

Sep 15, 2023 4 min read

Last weekend I gave a talk in a generative AI meetup about how we at Peerlist leveraging AI to improve the overall user experience of our product. This is kind of a written version of it.

What is Generative AI?

Generative AI refers to a type of artificial intelligence that can generate new content. Instead of just processing information or following specific commands, it can create something new, be it text, images, music, or other forms of content. It does this by training on vast amounts of data, understanding patterns, and then producing outputs that mimic the data it has seen. A popular example of generative AI is the Generative Adversarial Network (GAN), where two neural networks "compete" to produce and refine content, resulting in increasingly accurate or realistic outputs.

Why we’re using AI?

Why we are using AI
Well, why not?!

The answer is really very simple. Because everyone was using it or at least talking about it!

Joke apart! Let’s accept the fact that —

AI is going to change the way we build products and the way we consume products/services.

Since I started using ChatGPT for sentence rephrasing and grammar checks, I didn't feel the need to renew my Grammarly subscription. Sorry, Grammarly! We wanted to get started with building small features powered by AI but the real challenge was not to deviate too much from our original product roadmap. We did not want to build “AI-Generated Peerlist Profiles”.

After much thinking (precisely 27 minutes), we chose a focus: "Enhancing User Experience Using AI." The brief was straightforward: build something using AI that will improve and create a delightful experience for Peerlist members. So we started tinkering with OpanAI APIs and built a very small feature.

Tl;dr Generator

The very first thing that we built as a POC is tl;dr generator. Well, I know it sounds like a to-do app of AI projects, but hear me out.

Roughly 40% of posts on Peerlist Scroll feature external links, typically directing to blog articles or GitHub repositories. Imagine being able to get a short summary of that content. The idea was simple and easy to implement. We thought it was a cool idea to build and we had zero expectations from it. But to our surprise, when you read a summary, you’re more likely to engage with it. We are seeing a 70% surge in interactions with posts featuring our AI-generated tl;dr summaries.

Generative AI tl;dr Generator
How the generated tldr for a Github repo is shown on Peerlist Scroll

Sometimes it does generate some vague summary for the link. But we will be improving it ASAP. Kudos to Yogini for building this in a day!

Skill Recommendation

Now things are getting a little serious with this one. We’ve been working a lot on improving our onboarding. In the last 8 weeks, we have improved our landing page conversion by whooping 10x. Not 10%, but by 10x.

Even after this, we found that there is one screen where maximum users are dropping and abandoning the signup process. After investigation, the culprit was Profile Tag input.

When you are creating your Peerlist profile, you need to add skills, tools, and technology. We call it profile tags. It’s a mandatory field and you need to add a minimum of 3 of them. Many users used to add any irrelevant skills and complete the onboarding or just close the window with frustration. This was bad!

It was primarily because of 2 reasons (this is my assumption):

  • Memory Block: Ironically, when prompted to list their skills, many individuals (myself included) draw a blank. Despite proficiency in numerous technologies, languages, and frameworks, they elude me during form-filling.
  • Search Challenges: Users had to sift through an extensive list of skills to locate and select the relevant one. Navigating and adding from this vast array, one skill at a time, was both time-consuming and daunting.
Skill Recommendation
How Peerlist suggest skills

Nakshatra has written a thorough article explaining how he built skills recommendation algorithm. And if you want to give it a try, check your profile settings.

Job Description Generator

Many companies hire with Peerlist for engineering and designer jobs. If you ask any recruiter or hiring manager about the pain of a job posting, the most common answer will be writing a job description for a job post. If you’re a recruiter or hiring manager you can relate to this strongly.

To make this process easy, we will be introducing a simple Job Description Generator. Once you fill in some basic information about the role you’re hiring for, we will be able to help you generate a job description with a click of a button. Super simple, but super helpful! This removes a lot of friction from the workflow and improves the experience of job posting.

Generative AI Job description Generator
Generative AI Job description Generator

These are some examples guiding our exploration into the exciting territory of Generative AI. Our aim was straightforward: start modestly, understand how to incorporate AI into enhancing user experience and provide added value to users on Peerlist.

We’re working on some amazing people recommendation and job+candidate matching algorithms. Stay tuned for more details in the upcoming weeks.