case study

Graph-Powered Market Intelligence with Centrly

Centrly is a MIT-alumni founded startup that provides market intelligence for corporate strategy and development teams. Hear how tapping into an external Knowledge Graph accelerated their product development.

One of the most impressive things about Diffbot for me personally, is how open they are with not just the, the Knowledge Graph that they build, but also the tooling around it, for example, the Natural Language and the Enhance API. The fundamental benefit is that it's allowed us to accelerate our product development roadmap, probably by months.

- Jyotishka Biswas, Co-founder and Head of Product, Centrly


About Centrly

Centrly is an MIT-alumni founded, Forbes 30 under 30 company that is on a mission to accelerate ambitious technology to market by making the innovation process faster and more systematic. We met their impressive CEO, Katia Paramonova, who is a 3rd generation nuclear engineer who attended MIT at the age of 16, when she came to Diffbot’s Knowledge Graph meetup.

Born out of her firsthand experience helping corporate innovation teams, Centrly leverages knowledge graphs to make sense of the often complex market landscape of startups and cutting edge technologies to uncover partnership and investment opportunities. Because these emerging technology startups are new and often lack public financials, conventional market intelligence tools like Bloomberg or Capital IQ fail to serve this use case, and don’t provide nearly the amount of nuance to evaluate these opportunities.

Centrly looks beyond basic financials to B2B relationships they’ve formed in the market, including strategic partnerships, licensing agreements, and supply chain agreements. They aggregate into a network graph and evaluate companies across multiple dimensions: traction, technology, partnerships to provide a 360 degree view of a company.

Interview with Jyotishka Biswas, Head of Product at Centrly

DB: Hello! Please introduce yourself.

JB: I’m Jyotishka Biswas. I’m the Co-founder and Head of Product at Centrly.

DB: Tell me about Centrly. What does it do and what problems does it solve for customers?

JB: Our customers are teams at large enterprises who are looking to enter, or in some cases, establish markets that are forming around new technologies. So what we do is, use a Knowledge Graph to help them discover and evaluate potential partners to actually execute on their strategies in those spaces.

DB: And where did the idea for Centrly come from?

JB: Katia, our CEO started Centrly after seeing firsthand how difficult it was for stakeholders that are working on complementary technologies to really find out about each other, connect, and collaborate.

So what really drives us forward is this question of: what if we could surface those connection opportunities between those that are working on game-changing technologies and the organizations that have the resources to actually bring those technologies to market faster?

How much faster could we facilitate the clean energy transition or develop new vaccines and really solve some of the world’s biggest challenges through technology?

DB: In short, how does Diffbot help Centrly?

JB: So we use Diffbot’s Knowledge Graph to form a backbone of company information about the targets that our customers are interested in. So, rather than compiling a huge amount of information manually ourselves, we’re able to just pull in a baseline of company information directly from Diffbot, and use that as a backbone of data that we can provide to our customers.

DB: So what is a Knowledge Graph and why are Knowledge Graphs useful for market intelligence?

JB: As you can imagine when we’re working in markets that aren’t yet fully matured that are still forming, basic financial signals about a company aren’t necessarily sufficient to really paint a picture of what’s going on, to really understand what’s going on. We need to look deeper, and look at the actual relationships that are forming between companies and organizations in those spaces. E.g., What are some early customers of companies in those spaces? What are the acquisitions being made? What are the technology partnerships being developed?

It turns out that a knowledge graph is a really natural way to capture this information because not only does it allow us to aggregate really rich information about entities, but it also allows us to capture relationships between those entities in a really flexible way that is actually really friendly for doing pretty powerful analytics on that, that are hard to do without a flexible structure, like a graph.

DB: As a market intelligence platform, what is the value of integrating with an external knowledge graph?

JB: Yeah, so one of the hardest parts of working with knowledge graphs is actually constructing them, right? And one of the hardest challenges of that is actually being able to combine information that’s coming from disparate sources like articles or press releases, and actually linking them to a solid source of truth.

So, with Diffbot’s Knowledge Graph, what we get are two things: we get a solid foundation of company information–we get that source of truth that we’re looking for. And second we are able to use Diffbot’s tools for actually linking pieces of information to that knowledge graph. For example, they have this Enhance API where we can just put in a company name and a URL and get all this detailed information about that company.

DB: What are some of the common pain points when it comes to assessing external data providers?

JB: When we evaluate external data providers there are three fundamental things that we look for. The first is the just the quality of the data. How comprehensive is it, how complete, or how thorough and how detailed is it? The second is trustworthiness of that data, which is, can we look at the source of how the data was aggregated–do we implicitly trust the data provider as the first party source? And the last thing is how friendly the licensing model is. Obviously, we’re providing this data and insights from this data to our customers, so we need to be able to use it legally. Diffbot just checks all of those boxes for us.

DB: Did you look into other solutions prior to partnering with Diffbot, and if so what were they lacking?

JB: Yeah, we have evaluated multiple solutions and in almost every case either the data wasn’t comprehensive or detailed enough, or the licensing model wasn’t friendly, or they didn’t meet our product requirements of being able to pass through sources and trust to our customers.

And that’s where Diffbot’s strengths really work for us because they’re able to aggregate multiple sources of data and actually provide provenance for each of the facts in their Knowledge Graph. So then we can pass those sources through to our customers, and they can just click on a link and say, “Ok, it turns out this relationship that we really care about, came from this article.” So I can directly see the point of evidence in favor of that relationship.

DB: So, ultimately, why did you decide to work with Diffbot?

JB: I think the decision to work with Diffbot really came down to a call we had with Mike, where he pointed us to some integration strategies and really asked for product feedback. So I think, ultimately it was not just the accessibility of that technology, but just how easy it was to work with the team and see how much they cared about our success as a company. And really I just believed in Diffbot’s mission of structuring and making accessible this wealth of information from the public web.

DB: From a high level, what have the results been of the partnership?

JB: What using Diffbot allowed us to do is, rather than spend a ton of engineering effort–which for a startup like ours, is very expensive–collecting this baseline of company information and building a company database from scratch, we were able to instead lean on Diffbot to provide that information and focus on our unique value-add, which is layering the unique pieces of information and linking it to that backbone and then providing the rich insights on top of that, that our customers are really looking for.

One of our targets is to have every company or organization that we track in our database be linked to a very credible third party data source, and right now, Diffbot is the primary third party data source that we use because it’s really easy to trust and verify the data that we get from them. So, one of the things that we ended up building is a “Diffbot button” that, if we’re missing information that we need about a specific company, we can just press that button to get the information and it’s automatically propagated through our own database.

DB: Great, and in your opinion, what’s the biggest benefit of working with Diffbot?

JB: It really comes down to ease of use, right? Their APIs are so accessible, and it’s not just the data, but it’s the tooling around the data. One of the most impressive things about Diffbot for me personally, is how open they are with not just the Knowledge Graph that they built, but also the tooling around it. So for example, the Natural Language and the Enhance API. So, it’s really easy to integrate Diffbot in multiple ways, even though we primarily lean on one. So, the fundamental benefit is that it’s allowed us to accelerate our product development roadmap, probably by months.

DB: How has the experience been working with the team?

JB: Every conversation we’ve had with Mike has been really positive. He’s an incredible product person. He asks great questions. He always asks for feedback on every single call and then has really specific suggestions for features that we should look at that could be helpful. And, in some cases, even newer features that we weren’t aware of. And because he demonstrates this understanding of our needs, we take those suggestions seriously and we have built further integrations past our initial one based on his feedback.

DB: Finally, would you recommend Diffbot to others?

JB: We would absolutely recommend Diffbot. One of the most impressive things about the company is the variety of their offerings around this Knowledge Graph they are building. Anyone who wants to use a rich dataset as a baseline for their product would find value in their Knowledge Graph. But, even outside of that, they have a pretty powerful Natural Language processing API and powerful features around crawling web pages, and various tools that are incredibly valuable, even on their own.

More about Centrly