Paradox was born from a big bet: The future of hiring — and software — was going to be conversational.
Since then, we've been partnering with our clients to change the status quo. Which, inherently, is pretty hard to change. I mean, that’s kind of the whole point, right? If it was constantly changing, then it wouldn’t be the status quo. It would just be, well, the quo.
Elvis Ha, our VP of product, believes that even more change is on its way, and another big bet is in the works. We got a chance to chat with Elvis about that big bet, and how it might change how candidates interact with technology entirely. Here’s what he had to say.
How have recent AI developments affected the candidate experience?
Elvis Ha: AI and its various forms have been creeping into our daily lives long before ChatGPT exploded into the public consciousness a year ago.
Consumers have already gotten comfortable engaging with AI, from asking Siri or Alexa questions about the weather, to using auto-complete when they text. And more invisibly, map directions, quant stock trading, and social media algorithms have quietly been transforming for a while. Our watches even use AI for crash detection.
The same principle is true for the candidate experience. Since 2016, our clients have been able to distill and simplify their application process into an intuitive conversation. There’s no more forms or logins or waiting. And now in that long line of evolution, we have re-asserted a new gold standard: a natural conversational candidate experience using Generative AI. There are still job application experiences out there today that remind us of how far we’ve come in the past five years.
Think about it: We used to have to learn complex UI click paths to get things done. It was time consuming and tedious — but everyone just accepted it because it was how everything worked. With AI, we’ve moved to a fully conversational process. Which is not just better because it's less time consuming and tedious, but because it’s the original interface of getting things done — the most universal, accessible, earliest modality known to society.
Even before the first technology was ever invented, people had conversations. It’s in our wiring to use conversations to solve problems. And as we added more and more 3-letter acronym technologies to the enterprise tech stack, we kind of got away from that. But now we’re able to combine Large Language Model developments to return to our human desire to converse. Which isn’t just great for people who think about AI everyday like me, but for everyone.
People with disabilities —those who have trouble seeing or using a mouse — are able to interact with technology on all cylinders like never before. And for cohorts who didn’t grow up with technology, the conversational modality is much easier than learning flows of click paths. Honestly, I’ll be vulnerable enough to admit that there are frequently new tool UIs that confound me – and this is my job.
What do you mean by “modality?”
Think of a modality as the medium in which you convey or consume information.
For example, content that you can read would be categorized as “text modality”. A “visual modality”, which can be images or video, is content that you can look at. Audio is another one — content that you listen to. And the same content can be delivered in different modalities. You can read a physical book or listen to the audiobook. Or even watch the movie based on the book. It’s all up to what the individual prefers.
Modalities can be sent or received. Input-to-output. Paradox's bread and butter has traditionally been conversational: text-to-text. You text words to an AI assistant, and the AI assistant responds with more words.
In the future, though, experiences will become more flexible and multi-modal. Text-to-video. Video-to-text. Video-to-audio. Everything will be completely interchangeable, and open up a whole new set of big bets on how users interact with software.
Can we use multimodal experiences to improve recruiting right now?
We’re getting there. Our goal is to 10x the leverage and efficiency that our recruiters have, not trap them in new, shiny boondoggles. It’s important that we finetune this next leapfrog jump so that we release the best possible experience into market.
From the beginning, our vision was that the best possible experience was conversational — which is traditionally that text-to-text experience. Now, with the way that technology and AI are progressing, we must stay attuned and adapt to a consumer’s changing multimodal preferences and expectations. Because you know who uses enterprise software? The same people that use consumer software.
In the span of 6 months, impressive AI models like GPT 4o and Gemini have demonstrated the ability to translate text to image really well, and vice versa. OpenAI has been teasing glimpses of the Hollywood-disrupting Sora, which is their text-to-video generation tool. These are all really great in pushing the boundaries of what you can create from a text prompt, but you’re still just using words: “Generate me a video.” “Generate me an idea for this presentation.”
Multimodal encompasses “any modality” to “any modality”. You could show the AI an Expo marker org chart and receive vocalized coaching on how to approach hiring initiatives as a first-time recruiting manager. Or you could be the one speaking into your earbuds coaching the AI to iterate on a personalized internal mobility video.
With the way that technology and AI are progressing, we must stay attuned and adapt to a consumer’s changing multimodal preferences and expectations.
The key to all of this, eventually, is that multimodality unlocks the ability for everyone to work in the way that’s best for them. Text, visuals, audio. You can give an input that’s convenient for you and receive an output that you prefer. Different modalities cater to different people, and different circumstances.
In my personal life, I instinctively use OpenAI’s voice assistant product. I’m not Googling anymore; I just speak into my phone with a stream of consciousness and trust that GPT-4o will sort it out. I’ve used its trustworthy advice in lieu of a golf caddy (because who is going to caddy for me?), a professional color theory stylist (popular in Seoul, but unbeknownst to the United States), and so much more.
And you can immediately imagine some cool things if you extrapolate that kind of personalization across the entire hiring experience.
How will this all change the candidate experience?
All candidates are already predisposed to consuming media in different ways. Some people like to learn visually; some people like to read; some people like to listen.
Ultimately, the candidate experience needs to be tailored so that it converts the candidate. And personalization has always worked as a recruiting technique. We need to contextualize our message — and our delivery of that message — to make sure that individual candidate knows they’re being listened to. Which, with current automation, takes an overwhelming amount of hours and capital.
We are getting to a point, though, where we can see a new way to do things. Where we can automatically engage all candidates in the way that they want to be engaged, and serve them content that feels less spam-y and more like a hug.
For example, a candidate’s offer letter could be organized upfront with perks that they expressed they care about throughout the hiring process. Or they could receive a video of interview tips specifically tailored for questions they’ve already asked. We’re talking about a hyper-personal level of engagement — one that customizes how each candidate interacts with technology, and one that is more likely to convert them to applicants.
In any case, we know the candidate experience will become a lot more dynamic, while always protecting the organization’s brand. If you thought the last five years produced a much needed and refreshing lift in experience standards, you are going to love the next five.
