AI Demo Jam is a new event series organized by Dnipro VC together with Data Phoenix. On March 20, six AI startups presented their products live during the first AI Demo Jam. In this video, you'll see the demo from Divinia.

Divinia's NLP technology boosts business revenue by delivering personalized NLP search and authentic content at scale, including product videos highlighting authentic selling points from reviews.

Transcript

❗
Please note that the transcript is AI-generated and may have errors. 

Ji: I promise I will be fast. I think my product is pretty straightforward. But we'll see. As a founder, my name is Ji Fang. I'm the founder and the CEO of Divinia. And I've been working in the field of AI and NLP for over 20 years. So the recent advancement of AI is absolutely exciting for me. Because when I graduated in 2006, people didn't actually know exactly what NLP was. So very happy to be here. Thank you all. And I think one of the panelists also touched on this. I think one of the biggest impact of AI recently is actually it's drastically lower the cost of content creation. And when it works, it's really cool. You're making movie or stuff. But there is one problem. for businesses who need to deliver actual authentic content to their target clients, that doesn't really work yet. Because just how many of you, for any AI-generated content, you think you can just trust it? So nobody raised their hand. That's where we step in, because I'm a firm believer of authenticity. For business to convey any product information to the client, it got to be authentic. Imagine you are a car dealer. You have someone who asks for a video about the car you are selling. What's the compelling highlights? How the car look like? Can you generate some pictures? Can you just randomly make up some pros and cons? No. And that's where we do. We actually only deliver authentic content in terms of product information. We only use authentic data like reviews, verified reviews as well. 

Dmytro: One minute. 

Ji: One minute, okay. Let me jump in to just show you. That's what we do. We turn static, authentic, but static images and reviews into product videos. So I will quickly show you just basically this video. This video basically was generated using static images we create online. It can also be the car dealer's inventory photos. And then that story here was distilled using our NLP model, our own proprietary model. Actually only focused on the review data we automatically identified all of the cool things people talked about about this car and this is not a Chat GPT script even though we do have a generative model in generating the language but every single line is corresponding to authentic highlights that consumers actually mentioned in the review. So that's our product, that's how we differ. Thank you.

Dmytro: Thank you. Questions? 

Audience: Yeah, what's it for? It's obviously for producing more than car advertisements. What's the general sort of problem space that you're addressing? 

Ji: It actually applies to all e-commerce. So it's not just the cars. We, it applies to hotels. For anything a consumer would want to buy, anything that has reviews. That's data where we can analyze, we can actually pull out the true pros and the cons. 

Audience: So the significance of the reviews is that this is your authentic sort of data that you're producing? You're not going out looking at product catalogs or sales brochures. You have some way of going and scouring around, finding authentic reviews and then that produces the content for it. Now, within the content that you produce, the audio, is there some way to go back and for someone to say, ok, this particular claim came from this particular source?

Ji: Yes, we can actually show the exact review quotes. Actually, the same information, we can use it to make search, NLP search. This is a kind of search query that our search engine will produce the same. I'm using the same car domain as an example. If someone searches for sports cars with a great sound system, our NOP engine will analyze, and knowing that you care about whether it's sporty, whether it has a great sound system, and all the results return. It highlights. These are the high positives people talked about this car, and each of them, evidence is super important. So if you look at…

Audience: blogs, Google reviews,Yelp…

Ji: Exactly, cars.com, KBB. So we have a robust engine that actually crawls all those web reviews, images, anything, and puts them together. So it can make search content, it can make SEO, can make video content like what I just showed you, yeah.

Audience: Yeah, really cool. 

Ji: Thank you.