娇色导航

[This transcript was auto-generated] Hi, everybody, welcome to DEMO, the show where we have companies come in, and they show us their latest product features and new platforms. Today, I'm joined by David Seuss. He is the CEO of Northern Light. Welcome to the show. Well, thank you.

So tell me a little bit about what Northern Light is, I believe you guys are a market intelligence platform and explain a little bit about what that means. So we provide knowledge management for market research, market intelligence and competitive intelligence.

For large organizations that are complicated, have lots of products, lots of markets, they compete in, we aggregate content that is relevant to those use cases like business news, technology news, syndicated market research on government databases, websites that are relevant, and we create a content set that is specifically targeted toward the requirements of our clients.

Okay, so who is your product generally aimed at within a company if you if you're talking about a large enterprise?

So the immediate interface that we have is usually with the market intelligence or competitive intelligence staff, but then they serve large internal audiences of product managers, marketing managers, key account sales executives, product development, product development management, we average over 5000 users per customer. Wow.

And okay, so now you've got … how long have you been around doing this this many years, we had our 20th birthday party last fall in Boston Harbor. Congratulations. But so obviously, back then generative AI was not a thing yet.

So but now you've now you're adding generative AI features? Well, Northern Light pioneered the use of machine learning and search applications for enterprises in 2017 using techniques that were predated generative AI things like Latent Semantic Indexing.

But the new kid on the block is generative AI and it has elevated search to a whole new level, the search paradigm is not different because of generative AI.

And how would a company use this as opposed to just having someone go to Google to look up market information about a company?

Well, first of all, it's the vetted quality sources that the company probably pays for that were as opposed to the you know, the the wild web, where you get good things and bad things. So really curating the content is key. It's also secure. It's a secure enterprise application.

Everything that happens on the platform is private to the company. And we often include the content, for example, that the client themselves has developed their own internal market studies their own internal competitive intelligence analyses. And you can't do that with Google.

Alright, so let's jump into the demo and see what what you guys have done with some generative AI features. Sure. So this is a typical Northern Light, single point portal and a real implementation.

This will be done in the clients, colors, logos, they could name it anything they wanted, they could have any kind of graphics that they want. There are different sections here, there are dashboards for curated collections of content on topics.

Some of our clients have as many as 200 dashboards, most will have 20, or 30. There are certain there's such search functions for different types of research, like syndicated market research, or news or other types. There's personalization, there's collaboration, and in some content management functions.

Today, we're going to talk about search. And we're gonna talk about generative AI specifically, I'll start with a new search just to show you how it works. So let's say that I'm a product manager competing with Oracle in the database world.

And I want to know what partnerships has Oracle formed recently. So I might just ask a question or use cases question answering for competitive intelligence and market intelligence. So what partnerships has Oracle announced? Question mark, it's a question. And then I execute the query.

And the query is resolved with three steps. The first step is a traditional search result, which is what you see here. These are search result cards. You don't have a citation, I have a little snippet of tax, that's traditional search exactly is it's been happening since 1994.

When Lycos the first search engine deployed, okay. Then the second step is on the right hand side, it's the answer to the user's questions from each of the documents on the on the left hand side.

So Oracle has announced several significant partnerships including collaborations with Microsoft, MasterCard, and HSBC. And it goes on and describes it, you can see that's from the document, a document, data veil, and you can see the name of the document there this, all answers are cited and hotlinked.

That's very important. You can vet the source and you can click through to the document to consume it in its entirety. Here's the second one, they announced a partnership with Clara to establish the first of the major public cloud regions at Columbia. They announced a partnership with AMD.

And so it goes there are 20 hits on this page. Every one of them has an answer every One of them is responsive to the user's question, what partnerships are they formed? That's the second step, then we have the third step.

The third step is we summarize the answers into an executive summary on the left hand side. That's the executive summary. So that's the summary of the answers on the right hand side, you will see that sometimes there are multiple documents that cite the same event.

So they get consolidated together, it's more efficient to go through the executive summary. But again, every observation is cited and hotlinked.

So that the user can vet the source and especially so they can consume the full the full document, okay, when they get before you added those generative AI features, it would have just been the search results.

We think we do a good job on search results, or our relevance ranking is optimized for business, research, and important ways. So but but this changes it, this just changes at the user.

We, the user now really knows why these documents are on point, they know that this document answers my question with these facts, and that makes me really want to consume it. Now did you work with a like an open AI or another?

Okay, so we're using an open AI enterprise, I say we use the API of open AI for genAI. So we're using the the GTP 3.5 Turbo model.

And accuracy is really important, in this case for a lot of your clients, right, it's essential, and they're all of them are worried about hallucination. We're using retrieval augmented generation or AG for short.

So we're taking the text from these documents that we know are from vetted high quality sources to start with, and we are telling in our prompt to the model, only answer the question from this text, do not rely on your training data, which is you know, web data.

Take only the content we submit. And this works really well with news with syndicated market research with the internal research of our clients, with journal literature with conference presentations with corporate financial reports. We've used this same technique on all of those, and it works great.

And was this recently deployed? You know, how long? Yeah, did you kind of make this public for your clients? So we started making it available in the summer of last year.

Users, the first users got a hold of it in November of last year, and feedback has been good been overwhelmingly positive. Do you charge a premium for this? Or is this part of just the overall pricing. about a 10%? increase in price? Yeah, okay.

And then it's a subscription? Or do they pay outright monthly service? So our pricing model for the platform is a fixed annual fee, which can vary based on the number of content sources that the client includes, and whether or not they take certain optional features that we offer.

So it's a fixed annual fee. For enterprise wide, there are no user fees, there are no usage fees, there are no storage fees. It's a fixed number for unlimited use enterprise wide and for the IT folks that are watching, is there any kind of installation issues?

Is it just easy for users to get set up? You need a browser. And that's about it. And SSL because it has to be secure.

So we'll ask that we'll work with the IT departments of our clients to establish SSL, that's really the only burden on on the client. And today, frankly, that's trivially simple.

Every IT department knows how to set up SSL and you know, and where can people go for more information on you know, if they're interested in learning more about this product, NorthernLight.com. That would be start there. All right. All right. David, thanks for joining us and thank you.