Family Office Stories

Thomas Herchik explains how Orca uses AI and zero-knowledge encryption to help family offices manage complex legal entity data.

19 min Listen to the original Episode page
Family OfficesLegal Entity ManagementAI Data ExtractionZero-Knowledge EncryptionPrivate EquityComplianceData SecurityFintech

Summary

Thomas Herchik, founder of Orca, discusses the platform's role in legal entity management for family offices, which comprise two-thirds of their client base. He argues that entity management is a critical 'third pillar' alongside accounting, driven by increasing global legislation and private equity complexity. Herchik highlights Orca's use of AI to extract data from source documents, contrasting bottom-up ingestion with top-down summaries to identify ownership gaps. The company employs zero-knowledge encryption to ensure data security, positioning itself as a secure vault where the provider holds no keys. Herchik predicts a market shift toward clients owning their data while outsourcing advisory work, emphasizing the need to filter out document drafts to prevent AI errors.

Key takeaways

  1. 1
    Family offices represent approximately two-thirds of Orca's client base.

    Thomas Herchik · 2:58

Episode sections

  1. 0:09 – 2:25

    Introduction and Orca's Origin Story

    Lisa Cornwell introduces Thomas Herchik, founder of Orca. Herchik explains that Orca was founded nine years ago based on his personal experience managing complex legal and tax structures after marrying into a European family office, highlighting the inefficiency of manual data gathering.

  2. 2:25 – 3:38

    Orca's Market Position and Client Base

    Herchik describes Orca as a legal entity management system operating in 28 countries. He notes that family offices constitute approximately two-thirds of their client base, though they also serve operating companies, law firms, and private equity.

  3. 3:38 – 5:34

    The Criticality of Data Management for Family Offices

    Herchik argues that legal entity management is the 'third pillar' of a family office, alongside accounting and reporting. He cites increasing global legislation and the complexity of private equity subscriptions as drivers for the necessity of robust data management.

  4. 5:34 – 7:46

    Structural and Coordination Complexity

    The speaker defines two types of complexity: structural (legal hierarchy, share classes, voting rights) and coordination (sharing data with internal/external stakeholders). He attributes rising complexity to global mobility, diverse passports, and private market investments.

  5. 7:46 – 9:45

    AI-Driven Data Extraction and Efficiency

    Herchik explains that Orca extracts data from legal documents into structured formats using AI. He cites a case where a client onboarded 250 entities in five hours by uploading 350 documents, contrasting this with the manual processes of the past.

  6. 9:45 – 11:48

    Top-Down vs. Bottom-Up Data Ingestion

    The speaker distinguishes between top-down ingestion (using summary charts, which cannot prove ownership) and bottom-up ingestion (uploading source documents like share purchase agreements). He highlights how bottom-up approaches reveal gaps, such as a 5% discrepancy in ownership chains.

  7. 11:48 – 13:49

    Security and Zero-Trust Encryption

    Herchik discusses the sensitivity of data (wills, passports) and Orca's use of zero-knowledge encryption. He compares their model to a Swiss safety deposit box where the provider has no keys, ensuring only the client can access the data.

  8. 13:49 – 18:36

    The 'Own Your Data, Outsource the Work' Model

    Herchik predicts a shift from outsourcing both data and work to owning data while outsourcing advisory work. He emphasizes the importance of distinguishing 'signal' (final signed documents) from 'noise' (drafts) to ensure AI accuracy and avoid hallucinations.

Claims worth verifying

Orca was founded nine years ago.

Unidentified speaker (UNKNOWN) ·0:44

How to verify: Check Orca's corporate registration records or press releases from approximately 2026-2026; verify against the podcast release date of 2026-11-19. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Orca operates in 28 countries.

Thomas Herchik ·2:45

How to verify: Check Orca's official website or corporate filings for geographic presence; verify via company press releases or LinkedIn employee locations. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Family offices constitute about two-thirds of Orca's client base.

Unidentified speaker (UNKNOWN) ·2:58

How to verify: Check Orca's marketing materials, case studies, or investor presentations for client segmentation data. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Private equity subscription agreements are now as burdensome as opening a bank account used to be.

Thomas Herchik ·4:32

How to verify: This is a subjective comparison; verify by reviewing industry reports on KYC/AML compliance timelines for bank accounts vs. PE fund subscriptions. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Errors in UBO reports can lead to jail time.

Unidentified speaker (UNKNOWN) ·5:05

How to verify: Research local laws in major jurisdictions (e.g., UK, EU member states) regarding penalties for false UBO disclosures. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

A client onboarded 250 entities in five hours by uploading 350 documents.

Unidentified speaker (UNKNOWN) ·8:56

How to verify: Check Orca's case studies or marketing materials for specific efficiency metrics; verify if such a case study exists publicly. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Orca uses zero-knowledge encryption where the provider holds no keys.

Unidentified speaker (UNKNOWN) ·12:26

How to verify: Review Orca's technical whitepapers, security documentation, or third-party security audits (e.g., SOC 2 reports) for encryption architecture details. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

A client had 90,000 documents, of which roughly 8,000 were final signed versions.

Lisa Cornwell ·17:21

How to verify: Check Orca's case studies for specific client data volume examples; verify if this specific ratio is cited in public materials. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Orca has hundreds of clients across many countries.

Thomas Herchik ·18:11

How to verify: Check Orca's website or press releases for total client count; verify via industry databases like PitchBook or Crunchbase if available. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Orca has been doing legal entity management for 8-9 years.

Thomas Herchik ·2:45

How to verify: Verify against the podcast release date of 2026-11-19; check company founding date via SEC EDGAR (if public) or business registries. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Do your own research

  • Verify Orca's security certifications and zero-knowledge encryption implementation through their technical documentation or third-party audits.
  • Review the specific jurisdictions mentioned regarding UBO reporting penalties to understand local legal risks.
  • Check Orca's client list and case studies for independent verification of the 250-entity onboarding claim.
  • Assess the accuracy of AI extraction by comparing Orca's output against manual audits of a sample of legal documents.
  • Investigate the regulatory landscape for family office entity management in the 28 countries where Orca operates.
  • Look for independent reviews or competitor analysis of Orca's platform compared to traditional legal entity management software.
  • Confirm the definition and scope of 'bottom-up' vs 'top-down' data ingestion methodologies in the context of legal ownership verification.
  • Review Orca's data privacy policy to understand data ownership rights and exit procedures for clients.
Full transcriptexpand

0:09Lisa Cornwell: Hi everyone and welcome to another episode of Family Office Stories. My name is Lisa Cornwell, and I am joined today by Thomas Herchik, who is the founder of Orca. Thomas, welcome to the podcast.

0:22Thomas Herchik: Thank you so much for having me, Lisa.

0:24Lisa Cornwell: Back to the podcast. Yeah,

0:25Thomas Herchik: second time around.

0:26Lisa Cornwell: Indeed. So for those who didn't catch the first episode, um, maybe you could start by telling us what inspired you to create AUCA, what AUCA is, um, and how your background influenced the design of Orca.

0:38Thomas Herchik: Happily, yes. Um, so back in the day, it's about nine years old, our company. It was founded on a personal use case.

0:44Unidentified speaker (UNKNOWN): So

0:45Thomas Herchik: my wife, uh I have the great fortune of being married to someone who is a fifth-generation fifth generation member of a European family. business with a family office to go with it. And um when we got married and my wife relocated to Switzerland to live with me, it set off a lot of tax uh alarm bells. I

1:01Unidentified speaker (UNKNOWN): bet.

1:01Thomas Herchik: And this was uh a process in which I was involved to help uh gather information that we needed to share with a new lawyer in a new country and

1:09Unidentified speaker (UNKNOWN): a

1:09Thomas Herchik: new authority, new tax authority, and um was very enlightening for someone who'd worked in finance my entire life, I thought I was sort of adept to most um compliance and otherwise issues. Yeah. And this was very, very uh in-depth analysis of a very large legal structure. We needed to pull together a whole load of information and share it with someone. And I just saw throughout the process how incredibly inefficient each individual step was. We needed structures structure charts, we needed high-level sort of visuals so people could familiarize themselves with the structure. Yeah. You know, 30,000 feet. These were in PowerPoint and Visio and the like. We needed more granular detail that we were capturing in Excel, sort of EIN numbers, board, you know, positions, all the things like this. And then ultimately we needed the legal documents as well. And gathering that information was in itself a challenge. It was all over the place. Making sense of it ourselves, and then being able to relay it to a new counterparty who knew nothing about the family and was sort of getting up to speed, uh, all in a very, you know, high stakes environment, I say, trying to get it right, making sure, you know, fifty versus fifty point one percent being ginormous differences in

2:25Unidentified speaker (UNKNOWN): this

2:25Thomas Herchik: world. Um so yeah, Orca essentially was born out of that experience. Uh it is a legal entity management system. I didn't know that word existed when we started the business. We were just looking to organize information to support sort of tax, legal and compliance workflows, but have subsequently learned what we do is legal entity management. We've been doing it for eight years or nine years, I should say, in about twenty-eight 28 countries worldwide. Yeah. It's been wonderful. It's been really, really interesting. Uh huge inbound from lots of places in the world. I never thought we'd be doing business, but we

2:58Unidentified speaker (UNKNOWN): really

2:58Thomas Herchik: love to. And um family offices are the majority of our clients, about two thirds of our client place are single or multifamily offices.

3:06Lisa Cornwell: Yeah.

3:06Thomas Herchik: But we've also come to learn that so many people are struggling with legal tax and compliance. And a lot of people have complicated structures. So we have operating companies like BMW Group, we have um uh uh tax so law firms, yeah, trust firms, PE, VC charities. I mean, yeah, it's been really,

3:27Unidentified speaker (UNKNOWN): really wonderful.

3:30Thomas Herchik: 100%. And yeah, there's a lot of people, and the structures are a lot more complicated than I thought they would be when we started this.

3:38Unidentified speaker (UNKNOWN): And

3:38Lisa Cornwell: so I mean you started to talk about it there, but why do you think then that data management is such a critical issue for family offices and high net worth families today?

3:46Thomas Herchik: So I I will preface this with the fact that my lens really is the tax, legal and compliance data. Yeah.

3:52Unidentified speaker (UNKNOWN): And

3:52Thomas Herchik: obviously family offices, we like to sort of refer to entity management as the third pillar of a family office. Generally they'll have an accounting system that'll consolidate you know, their books of record a reporting system uh for the performance figures, et cetera. And legal entity management is our area. And why it's critical, um it is an absolute must-have. You know, no one is getting out of tax legal and compliance. True. Um the direction of travel is just that there's more and more of it. Um so it it is definitely not going away. Uh I think we've seen a lot of legislation also globally coming around that it is very clear this is going more in-depth. It's affecting more and more workflows. You know, private equity subscription agreements are what opening a bank account used to be like. And

4:38Unidentified speaker (UNKNOWN): opening

4:39Thomas Herchik: a bank account nowadays is, you know, I mean, you practically need dental records. Um definitely we would say it's a necessary evil. People ne are are are going to encounter this more. and more. At the same point in time, I said briefly earlier, it's we call it a high stakes game.

4:57Lisa Cornwell: Yeah. And

4:57Thomas Herchik: the reason we say that is, you know, if you get it wrong, there can be financial penalties. Uh in certain jurisdictions, there's a risk of jail time if you get something wrong

5:05Unidentified speaker (UNKNOWN): in

5:06Thomas Herchik: a ultimate beneficial owner report vis-a-vis government, for example.

5:09Lisa Cornwell: Yeah.

5:10Thomas Herchik: Um, and there's personal liability for the people involved. So this is something that needs to be taken seriously. It's not going anywhere. And there are solutions nowadays. It is you we no longer need to do what my wife and I did many, many years ago. The PowerPoint

5:25Unidentified speaker (UNKNOWN): Excel. Yeah, exactly.

5:27Thomas Herchik: Document, doc, Dropbox uh hack. Um, so for sure something that that can be also easily So fixed nowadays.

5:34Lisa Cornwell: Yeah. So maybe let's let's go into that a bit. Can you describe how Walker handles the complexity? of managing these multi-generational and multi entity data points that are coming in?

5:44Thomas Herchik: Yeah, yeah, really happily. Um so first and foremost, we like to think in complexity in sort of two terms. Um, so we'll refer to one as structural complexity, meaning how complicated is the legal structure? Are there, you know, five levels between the UBO and the ultimate asset that they hold? Are they using multiple share classes? Are they distinguishing between legal ownership, economic ownership, voting rights, non-voting rights?

6:09Unidentified speaker (UNKNOWN): Are

6:10Thomas Herchik: they holding control via board seats or

6:11Unidentified speaker (UNKNOWN): powers

6:12Thomas Herchik: of attorney? I mean, there's a million different things that can culminate in what we call structural complexity. And then the second element of complexity for us is coordination complexity, which is ultimately tax legal and compliance are team sports. sport, so to say. Yeah. You constantly need to share data with the outside world. Uh so both internal, I should say, stakeholders as well as external stakeholders.

6:35Lisa Cornwell: Yeah.

6:36Thomas Herchik: And those two elements, if if anyone has either of those, they have a complicated legal structure, or they have a structure that just a lot of people need to know about in different degrees of granularity, um, or they have, you know, the combination of both. that is what leads to the complexity. And I would say nowadays people are having more and more structural and coordination complexity through the nature of investing in private markets, investing globally, being more mobile, uh, you know, marrying people with different passports, moving

7:10Unidentified speaker (UNKNOWN): to

7:10Thomas Herchik: different jurisdictions. I think a lot of people are are fully embracing the freedoms that we have. nowadays and and are having to deal with the tax, legal, and compliance consequences as a result thereof. So those things certainly complicate it if you are marrying one or many people in different places, moving around, investing across a broad array of asset classes. And we see a lot in private markets, obviously, which are just paperwork from a paperwork perspective, much more intensive.

7:39Lisa Cornwell: Yeah. All

7:39Thomas Herchik: of that's uh combining into the the need to solve this. And you sort of asked how we're doing

7:45Lisa Cornwell: it.

7:46Thomas Herchik: Um entity management's quite simple at the end of the day. Our job is to map out all of the legal relationships that someone has. So it could be that they're on the board of a company, they own a certain number of shares, that they're married to someone, where's their legal address, their tax ID, all this sort of stuff. And every single element of what we're trying to map out originates in some version of a legal document.

8:10Lisa Cornwell: Yeah.

8:10Thomas Herchik: And essentially nowadays, what you're looking to do with entity management is just extract all of that data from those documents into a structured format that you can then leverage and you know use used for different things. And AI has played such a massive role in doing that for our company specifically. The ability to, you know, take raw documents and just extract all of the legalese from it is insane nowadays. I mean, going back 10 years ago, I wouldn't have uh assumed we would be where we are today. And

8:43Unidentified speaker (UNKNOWN): we

8:43Thomas Herchik: are very happy to be where we are today. I think the clients are thrilled to be where we are today. Um

8:49Unidentified speaker (UNKNOWN): And

8:49Thomas Herchik: it's gotten so easy We have examples of someone onboarding themselves in five hours, you know, 250 legal entities

8:56Unidentified speaker (UNKNOWN): up and

8:56Thomas Herchik: running in five hours.

8:57Unidentified speaker (UNKNOWN): Yeah.

8:58Thomas Herchik: Just having uploaded in the vicinity of 350 documents or got other examples of people who said, you know, previously I was just looking to understand the status quo. Now it's so easy to get data in. I'm going I'm going gonna to go back a hundred years. I want to know the full history of our family.

9:13Lisa Cornwell: Yeah. Um

9:14Thomas Herchik: and so that's been very, very liberating for us. Uh is it gay, it really gives people the ability to get so much more value out of the software, as well as for the clients, they can sort of, you know, with a click of a couple buttons, they are in a really meaningful place.

9:28Lisa Cornwell: Yeah, yeah. It's amazing. And I mean, we see all the time people struggling with exactly this problem because it's the first thing we ask if we're trying to do some analysis, you know. who owns this, where's it owned, how's it owned, what's your role? And very often it's hard to find that information.

9:45Thomas Herchik: We we see that a lot. And we're having com collaboration with a number of firms such as PWC in the onboarding process essentially. So it's you know this first touch point where you really are just trying to wrap your arms around the the family, so

9:58Unidentified speaker (UNKNOWN): to

9:58Thomas Herchik: say we need a basic understanding of are we talking you know, hundred 100 entities? or a thousand? Uh, we've seen the all of this. Um, and it's really interesting. This uh I might just go down a brief rabbit hole. Please

10:11Unidentified speaker (UNKNOWN): do.

10:11Thomas Herchik: We we sort of distinguish when we're talking about capturing data from clients in a top-down motion

10:18Unidentified speaker (UNKNOWN): or

10:18Thomas Herchik: a bottom-up.

10:19Unidentified speaker (UNKNOWN): And

10:19Thomas Herchik: the distinction between the two is basically if you gave me a structure chart now, which is uh, you know, let's say it's a hand-drawn uh diagram with boxes and lines and ownership percentages and and the like, it's essentially a summary of maybe a hundred legal documents that underpin that those boxes and lines. But the the visual itself is you know on one page and it's very simplified. So we can ingest data like that, and the system will extract everything that's on that page, but it ultimately can't prove any of that. It's just saying this is what we know so far about the person. And this is our top-down approach. So one document in, you already get quite a lot out of it. The bottom up is then you say, well, let's start uploading those share purchase agreements, the loan agreements, the operating agreements. And let's see once we start doing the bottoming up, how much of that picture can be reconstructed. So if the structured chart said this person owns 50% of this entity, and we've uploaded three share purchase agreements and it adds up to 45%, well now you have a five percent gap in

11:20Unidentified speaker (UNKNOWN): your

11:20Thomas Herchik: ownership chain. And this is really important for our clients to be able to prove ownership

11:27Lisa Cornwell: at

11:27Thomas Herchik: all points throughout time.

11:28Lisa Cornwell: Yeah.

11:29Thomas Herchik: And the AI just does a great job of sort of surfacing look, there's a little bit of an inconsistency here or this adds up to over 100% a hundred percent of ownership or whatever else it is.

11:38Lisa Cornwell: Yeah. And

11:38Thomas Herchik: it it helps them very quickly to identify if and where they have gaps in their own data, which we all hope we don't, but uh inevitably we we do.

11:47Lisa Cornwell: Yeah. So let's let's talk about something that also comes up, I think, with clients quite often is is how their data remains private and secure. So what are you doing at Orca to ensure that?

11:59Thomas Herchik: Yeah. I mean one of the most important topics hands down. Uh Um, especially because the information that we're consolidating will include things like their personal addresses, their wills, their passport copies, uh prenups, you know, all of the

12:12Unidentified speaker (UNKNOWN): like.

12:13Thomas Herchik: So it's very, very sensitive information. We like to say it's sort of the weapons grade plutonium. If it gets out,

12:18Unidentified speaker (UNKNOWN): you

12:18Thomas Herchik: know, th there's major consequences. Um, and we've taken an approach called zero trust encryption or zero

12:26Unidentified speaker (UNKNOWN): knowledge

12:26Thomas Herchik: encryption. And summarizing it in as few words as possible, I I like to say everybody nowadays will tell you the data is encrypted. And for me, there's one really helpful follow-on question, which is who has the keys? Because who

12:39Unidentified speaker (UNKNOWN): has

12:39Thomas Herchik: the keys determines, you know, if 10,000 keys have been created, well, I I wouldn't feel quite so secure as if, you know, there's only one or two.

12:47Unidentified speaker (UNKNOWN): And

12:47Thomas Herchik: standard practice is the client has a key to be able to decrypt the data. Standard practice also would be that the provider, so in this instance, let's say Orca would have a key to be able to decrypt it. And in our setup, we don't have a key. So we like to think of it sort of as a Swiss safety deposit box.

13:03Lisa Cornwell: Yeah.

13:03Thomas Herchik: We give the client the infrastructure to store data, but only they can open it and

13:08Unidentified speaker (UNKNOWN): access

13:09Thomas Herchik: That's it. So we know, you know, that they have a big box or a small box or a combination of boxes. Yeah.

13:14Lisa Cornwell: And

13:14Thomas Herchik: we'll see how frequently they come to visit those boxes, but we don't

13:18Unidentified speaker (UNKNOWN): know what they've

13:18Thomas Herchik: put in. We

13:19Unidentified speaker (UNKNOWN): don't

13:19Thomas Herchik: know what they've taken out. None of that information is privy to us. And I would say that is um that's really it gives you a really safe way of embracing the cloud. Um another thing that we're seeing a lot in the family office industry nowadays is people going back to on on-prem. prem.

13:35Lisa Cornwell: Yeah.

13:35Thomas Herchik: We're not quite doing that at the moment. It it would be quite easy for us. to do it technically, but it's definitely, I would say, a topic that's coming up a lot more. And especially with information of this degree of sensitivity.

13:47Lisa Cornwell: Yeah.

13:49Unidentified speaker (UNKNOWN): So

13:49Lisa Cornwell: we are coming to the end of our time together. Just a couple more questions. So how do you see this space evolving?

13:56Thomas Herchik: So I think uh it's specifically to entity management. I think it's going to just be so much. easier to adopt this. I think classically there were two hurdles to adopting entity management software. One was I have to go and gather all of my data,

14:11Unidentified speaker (UNKNOWN): which

14:11Thomas Herchik: uh every time we've gone through this is often in many, many places, more than than some would uh initially believe. You know, some is drawn up by that it's still at the law firm who drew up the initial contracting, some is with the business partner, some is in my inbox. I mean there's a a a multitude of sources that they'll be gathering this information from. Um

14:32Unidentified speaker (UNKNOWN): I

14:32Thomas Herchik: don't think that's going to change. But I think it's going to be easier to pull data together. I think a lot of firms are also more used to sending information over. It's going to be easier when you are gathering information to make sense of it fast and easily. That's already happening nowadays. And I think the premium for owning your own data is just going to get more and more and more, and it's going to incentivize more people to do that.

14:57Lisa Cornwell: Yeah.

14:58Thomas Herchik: So we see previously saw people sort of outsourcing the work. You know, it's better for office, for family office, let's say, to liaise with the best tax advisors across all of the best service, uh you know, law firms, tax firms in the market. versus trying to in-house everything.

15:16Lisa Cornwell: Yeah.

15:17Thomas Herchik: Classically speaking, they were outsourcing the work and outsourcing their data.

15:21Lisa Cornwell: Yeah.

15:22Thomas Herchik: We see them now owning their data and outsourcing the work. And I think that's 100% going to happen. And I think the benefit of owning your own data, given that it's easier now, it's also enabling you to adopt AI in a really interesting way.

15:39Lisa Cornwell: Yeah.

15:39Thomas Herchik: Um, so this is these are two sort of developments we we think are we're seeing in the market.

15:44Lisa Cornwell: Yeah. So so own your data, outsource the work. And is there any other piece of advice you give a family who's looking at this, you know, starting today?

15:52Thomas Herchik: Yeah, happily. Um so I would say owning your data, what does that mean for me? If I can point to a structure chart that says I own this, but I don't have the underlying document, I wouldn't consider that I own the data. I think you need to go down to the source documents.

16:08Lisa Cornwell: Yeah.

16:08Thomas Herchik: Because whenever there's a hesitation or a doubt, that's ultimately what you need to prove something. So I think what owning your data means is really important to understand. And I would encourage everybody to start from the documents up. Yeah. Um the other thing that I would say is w owning your data doesn't mean owning every single document that ever existed.

16:28Unidentified speaker (UNKNOWN): Um

16:29Thomas Herchik: we we draw a big distinction between work in progress documents. So let's say shareholder version shareholder agreements version one through seven, those

16:37Unidentified speaker (UNKNOWN): are all

16:37Thomas Herchik: the drafts, and then eight was committed to PDF and signed. You know, that for us is the relevant one.

16:43Lisa Cornwell: Yeah.

16:43Unidentified speaker (UNKNOWN): Uh

16:43Thomas Herchik: one through seven.

16:45Lisa Cornwell: Who cares?

16:45Thomas Herchik: Good yeah. Put it bluntly, absolutely. Yeah. Who cares? Um And so there's a distinction between what you can say signal and noise.

16:53Unidentified speaker (UNKNOWN): One

16:53Thomas Herchik: through seven are noise, eight is signal.

16:56Lisa Cornwell: Yeah.

16:56Thomas Herchik: Um, and I think that gives you fewer things to gather, makes it easier. And coming back to what we were saying earlier, now you have this sort of clean record of all of your legally binding final documents, and then you're extracting structured data from that. When you then feed a hundred percent signal to an AI. that's really meaningful. Yeah

17:17Lisa Cornwell: very powerful.

17:18Thomas Herchik: We've seen other clients, you know, we had a client who had 90,000 documents.

17:21Lisa Cornwell: Wow. And

17:22Unidentified speaker (UNKNOWN): they

17:22Thomas Herchik: said, you know, should we put all of them in Orca? And we said, heavens no, please just put the final legally binding versions in. It was roughly 8,000, you know, and that's a sort of one to ten ratio Yeah.

17:33Unidentified speaker (UNKNOWN): of

17:33Thomas Herchik: signal to noise.

17:35Lisa Cornwell: Yeah.

17:35Thomas Herchik: If you put 90,000 ninety thousand documents in an AI, you will get a lot of hallucinations because

17:40Unidentified speaker (UNKNOWN): of the

17:40Thomas Herchik: size of the data and you'll get a lot of answers that are based on noise potentially. I think it's very difficult for an AI sometimes to distinguish between signal and noise, and that's horrible. Yeah.

17:51Unidentified speaker (UNKNOWN): Uh

17:51Thomas Herchik: worst case scenario. Um and yeah, anything else for people just starting out this? Obviously they should check out with orca.com. I mean, we're we're doing this day and night. Uh we believe we've found really phenomenal solutions to it as evidenced through hundreds of clients across across a great number of countries and we're very passionate about this and above and beyond anything. We just love talking to people about it.

18:14Lisa Cornwell: Yeah yeah I think that seems a good place to close today's session. So Thomas thank you so much for joining us and thank you everyone for listening. See you next time.

18:23Thomas Herchik: Thanks Lisa.

18:24Lisa Cornwell: Bye

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