惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

Stack Overflow Blog
Stack Overflow Blog
PCI Perspectives
PCI Perspectives
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
让小产品的独立变现更简单 - ezindie.com
让小产品的独立变现更简单 - ezindie.com
V2EX - 技术
V2EX - 技术
Google DeepMind News
Google DeepMind News
量子位
博客园_首页
S
SegmentFault 最新的问题
S
Secure Thoughts
F
Full Disclosure
H
Hacker News: Front Page
博客园 - 三生石上(FineUI控件)
U
Unit 42
H
Heimdal Security Blog
N
News and Events Feed by Topic
A
About on SuperTechFans
C
CERT Recently Published Vulnerability Notes
Cyberwarzone
Cyberwarzone
Help Net Security
Help Net Security
The Hacker News
The Hacker News
L
LINUX DO - 最新话题
Application and Cybersecurity Blog
Application and Cybersecurity Blog
罗磊的独立博客
N
News | PayPal Newsroom
Spread Privacy
Spread Privacy
C
Cisco Blogs
C
CXSECURITY Database RSS Feed - CXSecurity.com
云风的 BLOG
云风的 BLOG
A
Arctic Wolf
Threat Intelligence Blog | Flashpoint
Threat Intelligence Blog | Flashpoint
Simon Willison's Weblog
Simon Willison's Weblog
B
Blog
人人都是产品经理
人人都是产品经理
TaoSecurity Blog
TaoSecurity Blog
博客园 - 【当耐特】
C
Cyber Attacks, Cyber Crime and Cyber Security
P
Proofpoint News Feed
Hugging Face - Blog
Hugging Face - Blog
I
InfoQ
D
DataBreaches.Net
大猫的无限游戏
大猫的无限游戏
Apple Machine Learning Research
Apple Machine Learning Research
L
LINUX DO - 热门话题
Google Online Security Blog
Google Online Security Blog
V
Visual Studio Blog
V
Vulnerabilities – Threatpost
Know Your Adversary
Know Your Adversary
CTFtime.org: upcoming CTF events
CTFtime.org: upcoming CTF events
B
Blog RSS Feed

product Archives – TechEmpower

Agentic Coding in Practice Product meets Engineering in the AI Era Using Generative AI to Drive Corporate Impact Selecting a Software Development Company in 2024 How to Use Generative AI and LLMs to Improve Search The Top 20 Symptoms of a Weak Development Team Technical Review: A Trusted Look Under the Hood Startup CTO or Developer Startup Metrics 53 Questions Developers Should Ask Innovators
Generative AI – The End of Empty Textboxes
Alan Laser · 2023-11-14 · via product Archives – TechEmpower

We recently completed a web-based application that uses a unique algorithm to match professionals with new career opportunities. As part of the onboarding process, the app asks both job seekers and employers what they’re looking for – in a text box – while providing a few suggestions in a pop-up.  If you’ve ever used a similar application, (or if you’ve ever used the Internet at all) you’ve probably seen this approach before.

And if you’re like most people, you’ve also probably struggled to fill those boxes in.  Everyone struggles with empty text boxes. Populating them can be hard work, especially when the content needs to be just right.  This isn’t just our opinion – our startup metrics prove it!  Even with the pop-up suggestions, we saw significant drop-off during user onboarding.  Drop-off on the first page of an application is bad news.  It means wasted advertising spend and lost goodwill.

The point is, empty textboxes aren’t just intimidating, they can significantly impact user engagement and conversion rates. On a different project, we’d just used a Large Language Model (LLM) – in this case OpenAI’s GPT – to provide users with pre-filled text boxes, with content based on choices they’d previously made.  Instead of an empty box, the user gets one filled with content to use as a jumping-off point.  If they like the content as-is, they can keep it; if not, they can change it.  Either way,  it’s a huge improvement over starting from scratch.

Applying that solution to our job matching site made perfect sense.  In fact, it makes sense for almost any application that has text boxes!  Leveraging LLMs to help users fill in text, whether it’s by providing starter content, samples, or highly personalized tips, makes their lives easier.  That’s why we’re planning on building most of our sites this way in the future.

Bottom line: with LLMs, empty text boxes are going away.

Profile Blurbs and Writing Prompts

Let’s look at our job matching site in more detail.  During signup, the professional is prompted to enter their profile into a form, with an upload box for a resume, fields for awards, skills and certifications, and then a textbox – 500 characters max – for their professional summary.

An empty textbox, demanding to be filled with a concise, compelling summary to impress potential employers is daunting. Fill it with the right words, and your dream job could be right around the corner. Get it wrong, and you’re just wasting fifty bucks a month.

For example, let’s consider Mark. He lives in Houston, he’s a dedicated math teacher, and a proud recipient of the Teach for America excellence award. He’s a right-brain guy, and not much of a writer, which makes an empty textbox a potential stumbling block. So instead, we fill it with a completely custom blurb, written just for him:

Hello! I’m Mark, an enthusiastic math teacher from Houston and a proud Teach for America honoree. I bring my passion for numbers to the classroom every day, drawing from my experience to inspire my students. My time with TFA instilled in me a deep commitment to education, inspiring me to dedicate my life to guiding young minds through the world of mathematics.

That blurb, and the following examples, were all generated from GPT in only a few seconds, at a cost of less than one penny. The information provided was all pulled from data he’s already entered – just Mark, Houston, Math Teacher, Teach for America.

And if this description doesn’t resonate with Mark, he can ask for a new one, while providing feedback.  Maybe those references to TFA sound like bragging, or he thinks “passion for numbers” sounds silly.  We could prompt Mark to enter descriptive keywords like “dedicated” and “engaging.”  This gives Mark more control over the process, without requiring him to write much, and gives the LLM more to work with.  Giving this feedback to GPT gives us a revised prompt – once again, in just a few seconds:

Hello! I’m Mark, a dedicated math teacher hailing from Houston. Every day, I channel my enthusiasm into creating exciting and engaging math lessons, leveraging my wealth of experience to motivate my students. My tenure at Teach For America cemented my commitment to education, motivating me to devote myself to navigating young minds through the captivating landscape of mathematics.

Thanks to his custom blurb, instead of closing his browser and feeling inadequate for the rest of the day, Mark completes his bio and becomes a happy customer.  He’s happy because the hardest thing he had to do all day was done for him – in a way that allowed him to maintain control of the output.  The LLM didn’t just write a blurb for Mark, it wrote the blurb that Mark would have written, if only he’d had the words.  Note how this version incorporates Mark’s feedback and descriptive keywords.  It’s this collaboration between the user and the LLM that drives good results.

In the pre-LLM era, an empty textbox was a tough challenge.  Now that AI is here to help, every textbox is an opportunity to help a customer through the challenge. Which means better conversions and happier customers. Happy customers mean happy CEOs and founders – and that’s what makes us happy.

One of the main powers of LLMs lies in their ability to generate text that not only makes sense but is also engaging and personalized. They learn from millions of pages of text, understanding patterns, context, and even nuances that make the generated blurbs feel almost human.

If having an LLM write the whole blurb seems like too much, we could also fill our empty textbox with a writing prompt:

I know writing about yourself can be tough. Start by introducing yourself, your profession, and your location. Share your daily drive, such as creating engaging math lessons, and how your experience aids this. Be sure to mention your Teach for America experience, and end by stating your mission, like guiding young minds through the world of mathematics.

Along with personalized tips that inspire creativity:

  • Inject Professionalism with a Touch of Fun: Mark, as a dedicated math teacher, you likely have engaging stories from the classroom. Share an anecdote that showcases your innovative teaching methods. This adds a touch of personality and makes your bio more memorable.
  • Be Genuine: Emphasize your genuine passion for teaching and your commitment to education. As a proud Houstonian and a teacher, highlight your connection to your community and your ongoing efforts to improve the world around you.
  • State Your Career Goals: You’re passionate about guiding young minds in mathematics. Clearly mention that you’re seeking a role where you can continue influencing students’ lives and making a difference.
  • Skills and Interests: Consider adding personal skills and interests.  Do you participate in professional development workshops, or perhaps enjoy problem-solving puzzles in your downtime? Including these details can provide a more holistic view of your professional persona.

Since LLMs can understand text as well as generate it, the tips could be interactive as well as personalized.  As Mark types, the text could be sent to GPT; if it determines that he’s added a teaching anecdote, a check mark would appear next to that tip.

Fill Every Textbox

Bio blurbs are a difficult sort of textbox to fill in.  But there are others almost as hard.  Think about text boxes that have stymied you in the past.  Look around your home or your office.  (If you’re remote, use your imagination.)  Do you see anyone frustrated by these empty textboxes? Off the top of my head, here are some examples, but the list is huge:

  1. E-commerce product descriptions and FAQs
  2. Real estate listings
  3. Social media posts
  4. Marketing emails
  5. Job postings
  6. Job application cover letters
  7. Personal statements for college applications
  8. Business proposals
  9. Customer support emails
  10. Press releases
  11. Dating profiles

All of these text boxes are difficult to fill in, though not necessarily for the same reason.  Product descriptions and listings require tedious editing to make them engaging – especially after the fortieth one. Customer support emails require both accurate information and a professional, helpful tone.  An LLM can help fill in all these boxes, either directly, or by providing prompts, tips, and editing help.

Are you a founder or CEO or head of product? Go through your product right now and look for empty text boxes. You might be surprised how many there are.  Each one represents a great opportunity to make your users more productive and happier.  A solid LLM integration can transform the way they interact with your platform. It can empower them to express their ideas more effectively and confidently, no matter what they’re writing.

TechEmpower can help

In the era of LLMs and Generative AI, empty textboxes are a product mistake. Instead, you can enhance the user experience and produce a better result by enabling the user with a combination of the right questions and a starting point. The result is less drop off, i.e., better conversion rates.

But there’s a big difference between an LLM implementation and the right implementation for you.  To get to what’s right for you, you need a tech partner with a deep understanding of your business needs, software development experience, data engineering skills and AI expertise.  With over 25 years of experience in the software industry – and many successful AI integrations under our belt – TechEmpower can help you replace those empty textboxes with happy customers.

Want to learn how TechEmpower can help you fill textboxes with GPT?