Marqo's Milestone: Securing $4.4M Funding for AI App Vector Search Engine

0:00Vector databases are a very crucial storage solution for unstructured data like images, videos, and text. And these are increasingly playing a very pivotal role in the AI landscape. Essentially, these databases power LLMs like GPT-4 that drive chat GPT. And cheaply, due to their kind of, you know, really powerful in real-time indexing and searching as data emerges or is updated, these functionalities are critical for tailoring features like personalization, recommendation systems, sentiment analysis, and beyond. Today on the podcast, I want to talk about a very interesting company in this space called Marco.

0:39This is an open source vector search engine for AI applications. So let's jump into it. With the escalating interest in generative AI, numerous different vector database startups have emerged as frontwriters. I think just in April, there was Pinecone and Weavatable that had significant investments over$100 million for Pinecone and$15 million for WeevyApp. And concurrently, newer entrants like Chroma and Quadrant aren't far behind. I think Chroma was$18 million and Quadrant was$7.5 million in their initial financing rounds. So further, I think the previous year, we saw Zillaz, which is the primary developer for the open source vector database Milvus and they secured an impressive$60 million in funding.

1:31And you know what, something that's really interesting is I was recently at the AI4 conference in Las Vegas. If you noticed last week, there was a couple days where I did not post a lot of podcast episodes. It's because I was up to my elbows in different conference talks. And while I was there, I spoke to a lot of the different AI companies that were on the market and a lot of the major funded AI companies at that conference were databases and people working on backends and architecture for artificial intelligence. So I think clearly there is a surging demand for companies developing infrastructure to kind of stay ahead of the AI revolution.

2:09And a standout among them, I believe, is the Australian startup Marco, which is creating a very comprehensive pathway to vector search. So Marco was conceived in Melbourne by its co-founders Jesse Clark, who previously headed machine learning at Amazon's robotic division, and also Tom Hammer, who is an ex-database software engineer for AWS, the primary challenge Marco addresses is the intricate world of unstructured data, which, as some sources suggest, compromises nearly 90 % of all data generated. So the growing reliance on generative AI for varied tasks underscores the imperative need for advanced tools to really navigate this massive data sphere that we're seeing today.

2:56So Marco sets itself apart by offering a very comprehensive suite of vector search capabilities, and they do this straight off of the shelf, which not a lot of other people are doing. So this encompasses vector generation, storage, and retrieval. And this integrate, essentially this is integrated, and they're doing this, you know, this approach to essentially allow users to use or get around third-party vector generation platforms from industry players like OpenAI, Hugging Faces, and really consolidating everything into a singular API, which is, if you know, if you've asked a lot of these developers working on this, is absolutely game-changing.

3:35So Tom Hammer, the CEO of Marco, he really emphasized the challenges developers face, stating that, quote, vector search is difficult to implement. Vector databases are only one part of the puzzle, and developers find it challenging to bring all the required components together to build a vector-based search experience with optimal relevance, latency, and reliability. Mocro provides an end-to-end system that brings all of these components together, solving a major pain point for developers. So I think, you know, moreover, Hammer kind of underlined the importance of the quality of search results.

4:11And he said, quote, if developers want to continually improve relevance of search results, they have to manually train new AI models for vector generation. Marco's continuous learning technology will allow search to automatically improve based on user engagement, which is absolutely crazy, right? So like, let's say, you know, they're integrating AI to make really powerful search results a company is, if they want the most up-to-date relevant information, they need a system that essentially is continuously improving. So it's not like whenever they go get around to doing a big new update or training a big new model and pushing it out, they get this one time, right?

4:50It's kind of like what we see with ChatGPT, right? It's like ChatGPT has these big pushes where it's like, okay, it's GPT-3, now it's GPT-3.5, now it's GPT-4. Imagine if it wasn't one big push for one big product, but it was just slow incremental improvement every single day. And that's kind of where they where they think this space should go for, you know, this this type of technology, some of the technologies they're working on. So talking about their funding last year, they actually raised an initial$840 ,000. And in a recent announcement, they just raised another $4.4 million in seed funding.

5:27And they really kind of with this new raise have hinted at expanded commercial endeavors. So I think this includes a new cloud service, which debuts alongside the pre-existing open source Marco project. And I think really embracing an open source model, Marco aims to resonate with the developer community, allowing them the freedom to really customize the product. So Hammer believes that open source development ensures superior product quality. However, he also acknowledges the resource intensive nature of maintaining open source projects, right? This is not something that's super easy. And, you know, leading to the introduction of Marco Cloud, which essentially eases the operational burden for users.

6:09So while rooted in Australia, Marco has established a parent company in the UK near its first investor, Creator Fund. And the company has a relatively modest presence in London, which it intends to augment, particularly in sales, marketing, and customer support roles to really fortify its European aspirations. So the recent funding round for Marco was spearheaded by Australian VC, which is Blackbird Ventures, with notable contributions from Creator Fund, January Capital, and Cohere co-founders Ivan Zhang and Adrian Gomez. And I think just having the co-founders of a company as prolific as Cohere on board with you really signals the value of your service, the value of the product they're offering.

6:52This is going to be a very interesting company to follow in the future and see how they grow.