RealTimes Studio

Mátyás Varga on Capila's deterministic data layer and AI strategy for real estate

12 min Listen to the original Episode page
Real Estate TechnologyArtificial IntelligenceData SecuritySaaSBusiness IntelligenceEnterprise SoftwareVenture CapitalStartup Strategy

Summary

Mátyás Varga, CEO of Capila (Cordia Intelligence), explains how his company uses a deterministic data layer to enable AI automation in real estate, addressing the industry's risk aversion. He notes that while IT sectors adopt AI easily, finance and real estate are harder to penetrate due to security concerns and outdated tools. Varga describes a sales strategy starting with free MVPs to demonstrate value, particularly in finance where chat-based answers are preferred over traditional BI reports. He highlights that IT security is the primary blocker, mitigated by using European data centers and strict access controls. Varga claims to have paying customers, including major Slovakian developers, and argues that the value lies in the data layer rather than the user interface.

Key takeaways

  1. 1
    security levels. So we no don't feed to the AI some security and some sensitive data. So we are creating security level, we are pre processing the data, and the processed data is just explained and basically presented by the AI.

    Mátyás Varga · 3:41

Episode sections

  1. 0:02 – 1:16

    Intro: Mátyás Varga and Capila

    Alexander Morari introduces Mátyás Varga, CEO and founder of Cordia Intelligence (Capila), which provides an AI platform for real estate solutions using a deterministic data layer.

  2. 1:16 – 2:36

    Why AI is hard to sell into real estate

    Morari notes AI implementation is difficult due to risk-averse decision-makers. Varga responds by starting with small, free MVPs to showcase value, especially in finance where chat-based answers are preferred over BI reports.

  3. 2:36 – 5:16

    Deterministic data and the AI layer

    Varga addresses LLM non-determinism by pre-processing data into deterministic tables before AI interaction, avoiding hallucinations. Security levels are implemented to filter sensitive data.

  4. 5:16 – 7:15

    Customers, dashboards and chat tools

    Varga states they have paying customers, including the largest real estate developers in Slovakia. They build modern dashboards from scratch and integrate AI chatbots into Microsoft Teams and Slack.

  5. 7:15 – 8:07

    Which sectors are open to AI

    Varga identifies IT companies as the easiest sector to penetrate, while finance, insurance, and banks are the hardest. He cites his background as a CFO for co-working spaces as relevant experience.

  6. 8:07 – 10:07

    Track record: from developer to co-working CFO

    Varga details past roles at HB Divis and HubHub, serving as CFO for over 10 co-working spaces in Europe. He notes real estate is outdated in tools and difficult to enter due to perceived high costs.

  7. 10:07 – 11:17

    IT security as the biggest blocker

    Varga identifies IT security as the primary obstacle, mitigated by using European data centers (e.g., Amazon) and strict security levels to overcome IT team objections.

  8. Morari concludes that AI will develop in real estate with IT and data security becoming priorities, and wishes Varga success.

Claims worth verifying

Varga is the CEO and founder of Cordia Intelligence, providing an AI platform for real estate solutions.

Unidentified speaker (UNKNOWN) ·0:24

How to verify: Check company registration records for 'Cordia Intelligence' or 'Capila' in Slovakia or Europe; verify Varga's LinkedIn profile. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.

The company starts with small MVPs, even for free, to showcase value to clients.

Mátyás Varga ·1:48

How to verify: Review case studies or press releases from the company detailing pilot programs or free trials. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.

The AI pre-processes data into deterministic tables before feeding it to the LLM to avoid hallucinations.

Mátyás Varga ·3:22

How to verify: Review technical whitepapers, documentation, or architecture diagrams published by the company. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.

The company uses multiple LLM providers and has a routing mechanism between them.

Mátyás Varga ·4:03

How to verify: Check API integrations or technical documentation for multi-model routing capabilities. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.

The company has paying customers, including the largest real estate developers in Slovakia.

Mátyás Varga ·5:16

How to verify: Check client lists on the company website or industry news for partnerships with major Slovak developers. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.

Varga worked as CFO for HubHub, a co-working platform, for two and a half years.

Unidentified speaker (UNKNOWN) ·8:18

How to verify: Check HubHub's historical press releases or Varga's LinkedIn profile for employment history. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.

Varga previously worked for HB Divis, a non-development company in Poland.

Unidentified speaker (UNKNOWN) ·8:21

How to verify: Check Varga's LinkedIn profile or company press releases for employment history. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.

The company uses Amazon data centers in Europe for hosting.

Unidentified speaker (UNKNOWN) ·10:39

How to verify: Check technical documentation or cloud infrastructure disclosures. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.

Do your own research

  • Verify Capila's customer list and revenue figures against public filings or press releases.
  • Check the specific security certifications of the European data centers used (e.g., AWS EU regions) for compliance with GDPR and industry standards.
  • Research the competitive landscape of AI-driven real estate platforms to assess Capila's differentiation.
  • Review the technical architecture of 'deterministic data layers' to understand the limitations compared to generative AI.
  • Investigate the churn rate and retention metrics of Capila's clients, particularly in the Slovakian market.
  • Analyze the cost structure of maintaining multi-LLM routing mechanisms versus single-provider models.
  • Look for case studies or whitepapers detailing the ROI of replacing traditional BI tools with Capila's dashboards.
  • Check for any conflicts of interest or partnerships with the LLM providers mentioned.
Full transcriptexpand
0:02
Alexander Morari: Alex Morari from Shelton Reed here. We continue with our series of executive peer reviews at Expo Real 2026, organized by RealTimes. And we have now the privilege of speaking to Mathias Varga from Cordia Intelligence. Mathias, before we double-click on the topics that you will present, introduce yourself, the
0:22
Alexander Morari: company, and what you guys are doing.
0:24
Unidentified speaker (UNKNOWN): Thank
0:24
Mátyás Varga: you for the invitation. Hi everybody. my name is Mathias Svarga. I am the CEO and the founder of the company Cordics Intelligence. We are providing an AI platform of real estate solutions. Uh basically to create in the data label over your data which AI can read, work with, and basically automate all of your processes over that.
0:44
Alexander Morari: Yes, and another then other important question questions. is
0:47
Alexander Morari: expo.
0:48
Mátyás Varga: Expo Expo sammy some very Apache. pache.
0:52
Mátyás Varga: Why?
0:54
Mátyás Varga: Maybe I will continue in English. I don't know.
0:58
Mátyás Varga: It is really nice. Basically, it is a lot of people, a lot of really interesting companies, and a lot of decision makers. So basically, here you can meet everybody who matters within the companies. So you don't need to organize a lot of meetings outside inside
1:13
Unidentified speaker (UNKNOWN): of your
1:13
Mátyás Varga: the city, but you can meet here and you will. can you can have a great experience with that.
1:17
Alexander Morari: Yeah, Mathias, you actually started a business which I think has a lot of uh stumbling blocks of or problems and issues. Why? Because AI implementation in real estate, I think, is one of those fortresses which is very difficult to break through. Decision makers in corporate that real estate contexts and landscapes are not very keen
1:37
Alexander Morari: and not very friendly. you could say, to AI. How do you manage to create not only create value on paper, but also implement AI solutions into your clients' businesses?
1:48
Mátyás Varga: Yeah. Exactly. It's not an easy, easy thing, but it's really interesting. So basically I'm enthusiastic about AI. I really like AI. And basically it was really hard to to get to start with work with some bigger fans, funds, really real estate funds, or development companies. But first of all, we always start with a small project.
2:08
Mátyás Varga: So we just say, okay, let's start with a small MVP, even for free. I'm okay with that one. And we can showcase you the value. And basically, what has the word has been changed? So, for instance, in finance, you don't need some BI report, Reports some big access, etc. et cetera, You need answers. If you can create a chat AI chatbot which can give you answers, you can talk with
2:28
Mátyás Varga: it, you can get all of the information that you need, then basically they are happy with the outcome and afterwards they are implementing all of these solutions. So
2:37
Alexander Morari: often, look, financial dimension, I think, in real estate business is a mission-critical thing where you cannot allow the AI tools or the whole agency agentec system, whatever you introduce, you can now allow them to be.
2:51
Alexander Morari: non-deterministic. And the answers if you chat to like regular chat boards or LLM-based kind of uh systems, they are non-deterministic. How do you resolve this issue?
3:02
Unidentified speaker (UNKNOWN): Yeah,
3:02
Mátyás Varga: basically what we are putting together is a data system. So basically we are creating a big database about your data, collecting all of your data. So basically, not just finance, but rent role, etc. And basically we are creating all our over top of them some deterministic views. So basically, we are creating some deterministic tables which we can feed to the
3:22
Mátyás Varga: LLM and to the AI. And the AI is just providing the explanation and some suggestions based on the deterministically created data. So that's what is the most important to understand that we are not processing the data via AI. It's it's not great because it's hallucinating. We are pro pre-processing pre processing the data. We are even creating different
3:41
Mátyás Varga: security levels. So we no don't feed to the AI some security and some sensitive data. So we are creating security level, we are pre processing the data, and the processed data is just explained and basically presented by the AI.
3:54
Alexander Morari: And you have data curation kind of or quality assurance loops as well?
3:58
Mátyás Varga: Yes.
3:59
Alexander Morari: Using the same LLM, LLMs so or the same tool, so you outsource to others.
4:03
Mátyás Varga: No, we yes, we outsource to others. So basically we are using a lot of LLMs, and basically we are even creating some controlling mechanism between the LLMs. So one LLM is controlling the other LM, etc. So we are using a lot of providers because in today's world, it's not like this is the best. This is the best. Basically, honestly, there is like five
4:23
Mátyás Varga: companies, and each week another LM album is the best. So you can can't even uh keep it up. Uh, I'm spending daily like three, four hours on RD to keep up with the face. But what we are saying is that we are using the top providers, the top models, and basically we are routing between them, and we have a great other for routing between these.
4:42
Alexander Morari: Yeah, exactly. It's like uh one of the influential YouTubers on AI says, uh AI never sleeps. So I understand you never sleep as well.
4:50
Mátyás Varga: As well. Yes, yes.
4:54
Alexander Morari: What is the biggest uh success so far you've had with real estate businesses implementing? So you said finances, yes. And you also have been also only piloting for now, or do you have a paying customer as well? Well
5:06
Mátyás Varga: we have paying customers, yes. So
5:07
Alexander Morari: can you disclose the name or describe the uh
5:10
Unidentified speaker (UNKNOWN): business sector?
5:11
Mátyás Varga: The name I don't want to disclose. discuss. I will I should have a look at to the NDA button. whether I can or not. Yeah,
5:15
Alexander Morari: of course.
5:16
Mátyás Varga: But we have the biggest uh real estate fast developers and set up in Slovakia. So in Slovakia we are quite well established. And
5:24
Mátyás Varga: for now, we have created then the financial dashboards, not just financials, but data dashboards, I would say. So basically they we are reconstructing Deconstrating all of the data, financial and non-financial, then we are providing a nice uh application for. them that is a dashboard, modern one, not BI tools, because BI2s are totally outdated. So we are building them
5:44
Mátyás Varga: from scratch, uh modern uh solutions. And on top of that, uh we always, first of all, we put together the data, then we discuss with them what data we can show to the LLL, to the AI. And this is AI chatbot you just implement into the chat platforms, Microsoft Teams, Slack, uh, et etc. cetera. So it's quite natural with them for them
6:04
Mátyás Varga: that they can talk with them just like chatting with a colleague or somebody else. You can chat with your data, with your AI tool, and it can combine with you everything. Presentations, create Excel, etc. So our goal is so they should not go to any SaaS SAS application. They just should talk with
6:24
Mátyás Varga: their with their their AI agents.
6:29
Unidentified speaker (UNKNOWN): And
6:29
Alexander Morari: eventually bycode something based on their own proprietary data
6:32
Unidentified speaker (UNKNOWN): systems.
6:33
Mátyás Varga: They can bycode anything, but it's not the core core thing which we are doing. So basically putting together the data, have the data uh level is the core
6:41
Unidentified speaker (UNKNOWN): uh
6:41
Mátyás Varga: business model. And to bycoding some front ends, it's just
6:45
Alexander Morari: Yeah, yeah.
6:45
Mátyás Varga: in one week you we can bycode byput any front end. So it's not an obstacle right now, and you should not think about it that I should do this big SAS tool, I don't think that's a correct way to think right now. I think SAS is dead, sorry, but that there's the future.
6:59
Unidentified speaker (UNKNOWN): Yeah,
7:00
Alexander Morari: exactly.
7:00
Mátyás Varga: And and you should fit a solution for you. So basically if you have some processes, you don't need to change the processes to fit the tool. The tool Two should fit your processes. So basically we can bycode wipe code any front end in in matter of weeks. It's not a an obstacle.
7:16
Alexander Morari: Exactly. Is there any preference in terms of asset classes that you see for your company development? Which is more open-minded? Yeah, or which potentially is easier to to should I say penetrate? Or let's say
7:29
Unidentified speaker (UNKNOWN): start
7:30
Alexander Morari: copyright.
7:32
Mátyás Varga: Yes,
7:33
Alexander Morari: exactly.
7:33
Mátyás Varga: It's it's really hard. Which is the easiest, easy starting I think the IT companies. So we are working with IT development companies. It is they are just thinking the same as me.
7:43
Mátyás Varga: And but real estate is not so hard as the finance is the hardest. So the finest insurance, etc. banks, there we have no track with them. But the easiest are the IT companies, I would say, and An ID ideal solution companies.
8:01
Unidentified speaker (UNKNOWN): So do
8:02
Alexander Morari: I understand correctly? you want to focus on real estate industry as such
8:05
Unidentified speaker (UNKNOWN): and provide solutions for
8:06
Alexander Morari: in real estate. My
8:07
Mátyás Varga: track record is from real estate. So basically, I used to work for the development company, then I was uh a CFO for uh co-working spaces, like 10 plus co-working spaces in Europe. And basically
8:18
Unidentified speaker (UNKNOWN): You're
8:18
Alexander Morari: avoiding mentioning the brands why?
8:20
Unidentified speaker (UNKNOWN): Well
8:20
Mátyás Varga: I can mention them. Mentioned that.
8:21
Unidentified speaker (UNKNOWN): Yeah, sorry. I
8:22
Mátyás Varga: was I I do know that it is okay. So I I used to work for HP HB Divis, Levis, uh which is a non-development company in Poland as
8:28
Unidentified speaker (UNKNOWN): well, I
8:29
Mátyás Varga: think. And afterwards HubHub, which is their uh co-working platform. I used to be the CFO there for two and a half years. And I really like real estate. By the way, it's a nice, nice uh thing. You are building uh we can see what what the real estate is building, and I like that. And I think real estate is really outdated on the
8:48
Mátyás Varga: tool side. So that's why. I have decided okay, we somehow need to get to the real estate. And right now we are working at the biggest fund in Slovakia, having the biggest developer in Slovakia, etc. So we are making progress on that one, but it's not easy to enter the market because everybody is thinking that okay, new IT thing to implement is three years budget
9:08
Mátyás Varga: of 20 twenty million,
9:10
Alexander Morari: Yeah.
9:10
Mátyás Varga: True. which is
9:11
Unidentified speaker (UNKNOWN): not good today.
9:12
Mátyás Varga: So today it can be in weeks and just uh a frame. fraction of the budget which is
9:16
Alexander Morari: Yeah, in your company in your company you want to focus on l on f only on financial dimension or would you venture into others, like marketing or leasing, like like CRM of new generation, you could say?
9:33
Mátyás Varga: Because it's always the dig biggest junk for the real estate companies, I think. But uh for these companies the leasing we are we are putting together the rent roles, basically, uh to extra data from lease agreements, to talk with lease agreements, marketing as well. So basically, our goal is to automate all of your company, to put all the data of your company into one place and
9:53
Mátyás Varga: then connect this data because one data needs to be connected to other data, et cetera, even to market data. So basically it's not just the data from your company, but to provide market. researches and sufferens.
10:04
Unidentified speaker (UNKNOWN): Yes,
10:04
Alexander Morari: external as well, yes. Okay. Let's imagine you are your dream customer, like one of the biggest developers in Europe, and so
10:12
Unidentified speaker (UNKNOWN): on.
10:13
Alexander Morari: What would be the most challenging question you would ask? You're seeing like what is missing in your company?
10:18
Unidentified speaker (UNKNOWN): Uh
10:19
Mátyás Varga: I think it's not missing in my company, but it's missing the understanding of the whole as a real estate student is the security. We are dealing a lot with IP. IT security questions, and that's the biggest threat for them that okay, AI is going to see my data, somebody else can prompt it so they can see that.
10:34
Alexander Morari: Some data center in the US or
10:37
Unidentified speaker (UNKNOWN): God
10:37
Alexander Morari: forbid in Asia and
10:39
Unidentified speaker (UNKNOWN): so
10:39
Alexander Morari: on.
10:39
Mátyás Varga: And for that one, we are we have answers. So we have a lot of answers. It can be treated. So we have data centers in uh in Europe, from Amazon. Uh it can be treated uh on different security levels, etc. So it was a big focus for us for the last three months, I would say, that we have been really focusing on the I2 security because it was a big blocker. So even
10:59
Mátyás Varga: if uh we had a presentation with some CFO or some C-level person, they liked the idea and they just dropped that, okay. My IT team needs to verify it.
11:08
Alexander Morari: Yeah.
11:08
Mátyás Varga: And with the IT team, it was like months and months discussions and it was really hard. So basically, this that's is the biggest. biggest obstacle, I think.
11:17
Unidentified speaker (UNKNOWN): Yeah,
11:18
Alexander Morari: so Matthias
11:20
Alexander Morari: just showed that AI will be developing in real estate industry for sure. Somewhere down the stream, IT security and data security is a priority uh issue. And we wish Matthias Matis, you would not crack the nuts of IT security and grow your company to the best. benefit of your
11:39
Alexander Morari: business partners, but overall also to the benefit of the s real estate industry as such. Thank you very much. That's a wrap.
11:46
Unidentified speaker (UNKNOWN): Thank you very
11:47
Mátyás Varga: much.
11:48
Unidentified speaker (UNKNOWN): Thank
11:48
Mátyás Varga: you. It was nice to meet you. And take
11:51
Unidentified speaker (UNKNOWN): a
This page was produced automatically: the audio was transcribed with nvidia/parakeet-tdt-0.6b-v3+sortformer-v2.1 and interpreted by Qwen3.6-35B-A3B-UD-IQ3_S.gguf. Transcription and interpretation can contain errors; always confirm against the original episode. Nothing here is investment advice.
All insights