Keewano DB, a Database Built for Machine Reasoning at Scale

Hey, everyone. Welcome back here to Techstrong TV. My next guest is Mark Gorodashov.

Mark is the CEO and co-founder of a company we're going to talk to you about, named Qwano. But before we talk about Qwano, I thought we'd talk about Mark a little bit and get a little bit of his background. Hey, Mark.

It's nice to have you here on Techstrong TV. Welcome. Good to be here.

So Mark, as I said, let's talk about you at first. Our audience always likes to know who they're watching and listening to. Before you were CEO and co-founder here at Qwano, tell us a little bit about your journey.

Oh, great. So back then, 20 years ago, I started my career as a software developer, and I led a small DBA team. So I worked for five, six years as a team lead at a DBA team.

Back then, we mostly worked with SQL, which was very popular. Sure. Microsoft SQL.

From there and going on, I founded two companies. This is my third company. Very good.

And yeah, I'm third time founder. Were the other two companies also in the database space? No, the first company was called Test Project.

It was in test automation at the cloud, like infrastructure for test automation. It was sold to Tricentis, an Insight Partners company. Sure.

And the second company was in the integration area with more or less 300 people in several R&D centers, large R&D centers, and we did large integration projects for the government, for big companies like Microsoft, Nvidia, et cetera. Beautiful. So what was the passion?

What was the story with Qwano? When did you found it? You didn't wake up one day and say, "Hey, I want to do this," right?

Tell us the genesis, if you will, of Qwano. Right. So about two and a half years ago, our friends from Plarium, Plarium is a big gaming company.

They're running multi-billion in revenue, and then they a high-valuation company in the gaming world. So my friend, the CTO from Plarium, and the general manager, come to us. Back then, me and my partner was on an L period.

We just, after selling a successful company, second company, and was kind of bored, in the enterprise after sell. And they came to us and said, "Hey, guys, we have something huge. Machines are becoming big consumer of data.

Today's databases wasn't built for the scale that's needed, and data warehouses, they're built for different purposes, for humans, for dashboards, for BI. " And back then, agents was mostly scientific. It's crazy how much changed in the two and a half years.

And yeah, I saw the idea, and as an ex-database expert, I was in love with the solution that they come with. At the beginning, it was only working on a laptop. On the CTO's laptop, he's just showing me the demo, written in pure C.

And I saw the engine, and I said to my partner, "We should do it for a third time. " Sure. Absolutely.

So after all these years, kind of returning to your database roots as well. Right. So there's some symmetry there, right?

Yeah. Fate. And it is an interesting time for this.

But the tagline is, Qwano DB, the first database built for machine reasoning at scale. Right. Let's take that apart.

Right? Yeah. What do you mean by that?

So, we realized that agents and machines need something else from what humans need. And it begins from scale and how much concurrent queries they can run and what kind of resolution they need. And it ends with context because they have their limitation.

They have a limitation in tokens and in context, how much they can process efficiently. So we realized that to try to answer those needs, instead of building some middlewares on existing data infrastructure, we understand that something fundamental need to be changed. And we come with a new storage format that build around how machine reason at scale.

And we started from there. So we built a new data infrastructure, a new database from scratch, from the storage to the engine, to the interfaces. All the way through.

And though your friend comes from the gaming industry, and so much innovation comes out of the gaming industry. Because of scale and the needs of it. Who's this really aimed at?

So I think that it's aimed for all databases, not the transactional databases. A database that need to store a lot of data. So whoever need to analyze billions of events at scale and need to get live results, not data that's aggregated and come weeks after or day after, but they need to reason now on what's happening.

And those who are building agents specifically on this data infrastructure, I think that's aimed for everyone. It can be fintech, it can be e-commerce, it can be product, it can be everything that come with B2C. If you have large scales of events.

So everything that need the volume and the speed that need it in the agentic era can benefit dramatically from our database. What about, let's say observability, right? Yeah, I knew it's a good question.

I knew you were going to raise it. Yeah. Okay.

So I think that it's very hot today, and it's very popular. Everyone's running after observability, and they was doing observability for servers, and now instead, everyone's doing observability for AI. But I think that observability, it's a big market that a lot of companies compete there today.

And I think that it also can be great for observability of agents, but for the behavior of the agent. Because observability have two parts. The first one is the behavior, the events, what happened, which tool was called, what sequence called to agent to hallucinate.

So it says the behavioral part. And the second part of observability come from logs. So all those prompts that running from the different tools, et cetera, and you also need to save the prompts themself.

So observability, it's a bigger term, and there are many things inside. So in terms of behavioral observability of agents, you can utilize Q1 ODB at scale. But in terms of the logs and textual parts, this is something that we do not support at the moment.

We're probably going to release it at next quarter or two. Got it. We've come a long way since SQL databases.

Probably the biggest thing back then was we went from SQL, and all of a sudden, we saw NoSQL databases. Then helped with scalability and speed. We've seen things like time series databases, graph databases.

Each new wave of database is trying to Be more scalable, be faster, be easier, more distributed, et cetera. Where do you think Keewano fits in that? I think that Keewano is a purposely built database for agents to reason over live data and over real data.

It's not like we are using a contact graph, a proximity or similarity or vector DBs. No. Basically, you give the agent the capability to access all of your agentic data warehouse without the predefined questions, because you don't know what agent is going to ask next, right?

So I saw several interviews when people were saying, "Yeah, we're going to optimize this clickhouse for something. We're going to optimize this index for something else," and they're optimizing the databases for predefined questions. But you don't have a clue what the agent is going to ask, or what type of connections or relations the agent is looking for over real data, not over something that was indexed or something that was processed.

So I think that Keewano DB helps in that specific area. So you have a capability to find connections and relations on different entities that you wasn't think on them in advance, but they might be very relevant to agent query or agent investigation or agent whatever. Mm-hmm.

Excellent. All right. You launched it.

Is it in general availability now? How do people download it, use it? What's the road here?

Yeah, we launched it next... Last week, sorry. And, it was exciting.

It's available on general availability, so if you want to write, connect your agent and, it's really easy. You can connect Cloud, you can connect any agent and start working with that. So yeah, and since we launched, I was approached by several large agent companies, I don't want to mention their names right now, but, for potential partnerships, and they want to evaluate their agents.

They are T1 companies, how the agents work on, they're curious how to see how the agents work on Keewano DB and what kind of benefits they give on top that other data infrastructure doesn't give. So we're in talks with them. Many downloads.

We have a webinar that's coming tomorrow. I invite everyone, if the people- Well, by the time, Mark, remember, the time this is edited, it may be past tomorrow. Oh.

Is it available on demand as well? Yeah, it will be available on demand. And we plan to share these webinars once in a week or twice a month or something like that.

So please join. Excellent. I didn't mention that, but, before we launched, we was working with design partners for two years- Mm-hmm ...

and for big companies, that generates billions or trillions of events per day, and those are the scales that we're speaking about when you speak on machine load. Yeah. And we didn't release it immediately.

We work with them, we learn what kind of feedbacks and what kind of needs they have. And surprisingly, they also have a need, like, everyone working with MCP, which is practically the standard today, but they also, like the data engineering team and other teams also want to get an access to the raw data. But how you give access to the raw data, where the raw data built for machines, right?

So we managed to build a simple SQL query. So now you can also- Really? Hmm?

Just a simple, plain old SQL? Yeah. So now if you have, I don't know, multi-billion scale of events, you can just run a simple query on that with affordable price and get answers in milliseconds.

So the power that the agents have right now, the data engineers, our design partners, and the first customers also utilize this power with a simple PostgreSQL. So we managed to build this layer of, let's say, standards or industry standards, protocol for those who wish to utilize it at scale as well, like a standard query. Which is fascinating me as like an ex-DBA.

Yeah, that is. That's almost too simple. Yeah, it's crazy.

I was running inner join or select on query, and the results was coming immediately and was like, wow. It was fun and enjoyable for me. Absolutely.

We never mentioned the website. What's the website's name? What's the URL for the website?

com. K-E-E-W-A-N-O, just the way it's going to be on the screen here. And Mark, lastly, how's this sold?

How do you make money here? How do we make money? Yeah.

What's the business model here? So the business model is, we have the small companies, and we encourage the adoption, the bottom up as well. So everyone can register to Keewano, even if you have a small budget.

We don't charge per event. So the first product team, it's kind of new in the product world at least. So they can send as many events they want per entity, and we were going to charge only per entity.

So it's very good. They can send as many events as they want. And we're also in talks with enterprises.

We have a team, a net team and US team that work with enterprises and partnerships to sell to them. They have different needs. Obviously, they want to work with Keewano aside their existing data warehouses.

They don't want to replace anything. So Keewano work perfectly aside existing data warehouse that they have. And some of them using us as a T1 agent framework, to reduce the costs and make the answers and the reasoning more accurate as well.

So for those enterprises, we have teams that working with them. Excellent. com.

Yeah? Yeah. Well, it's all there available.

Right now. Yeah. Hey, Mark, we're about out of time.

I want to thank you for coming on here and telling us all about Keewano. We wish you much success with it. Come back and keep us posted, okay?

Thank you so much. I will wait for your invitation for the next invite. Don't wait.

Never wait. Just let us know when you've got something to tell us. Will do.

All right? All right. Yeah.

Keewano. KeewanoDB. com.

We're going to take a break here on Techstrong TV. We'll be right back.