All Things Fundraising

Taylor Lowe argues private capital firms must fix fragmented data foundations before implementing AI workflows.

35 min Listen to the original
Private MarketsArtificial IntelligenceData InfrastructureCRMVenture CapitalFintech

Summary

Taylor Lowe, CEO of Metal, argues that private capital firms fail at AI adoption because they prioritize workflow automation over data infrastructure. He claims investment firms suffer from fragmented, unstructured data stored across CRMs and emails, which LLMs can normalize into a usable foundation. Lowe criticizes traditional CRMs like Salesforce for user-induced data decay and advocates for a 'headless' CRM approach using automated agents to ensure data fidelity. He asserts that generic LLMs lack the nuance for private market documents, requiring explicit classification and provenance tracking. The episode concludes that a robust data foundation reduces vendor surface area and enables effective benchmarking against external sources.

Key takeaways

  1. 1
    Metal is a New York-based AI company building a 'context layer' for private capital, integrating deal history, relationships, and firm judgment into AI tools.

    Unidentified speaker (speaker_0) · 1:33

  2. 2
    Investment firms lack a ubiquitous source of record; their most valuable asset is internal data stored in fragmented, unstructured formats across CRMs, emails, and file repositories.

    Unidentified speaker (speaker_2) · 7:49

  3. 3
    every time you bring in a piece of data into your database, you poison your database. Right. And why is that? Because it starts to decay day one.

    Unidentified speaker (speaker_0) · 26:11

  4. 4
    Benchmarking is inefficient without normalized data, as it requires manual reconciliation across six to seven sources and multiple stakeholders.

    Unidentified speaker (speaker_2) · 30:23

Episode sections

  1. 0:10 – 1:23

    Podcast Intro and Sponsor Read

    Host Gui Costin introduces the Rainmaker Podcast and a sponsor read for Dakota Marketplace, a database for fundraisers.

  2. 1:23 – 4:40

    Guest Introduction and Background

    Taylor Lowe, CEO of Metal, outlines his career from SDR at Copper to co-founding Metal with Y Combinator in early 2023.

  3. 4:40 – 6:50

    YC Frameworks and Market Truth

    Lowe discusses the Y Combinator experience, emphasizing filtering out hype to focus on market truths and Metal's pivot from horizontal LLM infrastructure to private markets.

  4. 6:50 – 12:08

    The Data Fragmentation Problem

    Speakers discuss the challenge of fragmented internal data in CRMs, file repositories, and emails, and how LLMs can normalize this unstructured data.

  5. 12:08 – 14:09

    LLMs as Components, Not the Whole System

    Lowe argues LLMs are just one component; Metal remains agnostic to model providers, allowing clients to access contextualized data via MCP.

  6. 14:09 – 19:30

    Data Foundation vs. Workflow

    Lowe critiques starting with workflow automation before solving data infrastructure, using a PowerPoint analogy to illustrate the need for foundational structure.

  7. 19:30 – 23:53

    Trust and Data Fidelity

    The conversation covers how Metal exposes data trust issues, often revealing input capture problems rather than AI hallucinations, creating a compounding flywheel effect.

  8. 23:53 – 24:48

    Reducing Vendor Surface Area

    Lowe concludes that a successful data strategy can reduce the number of tools a firm uses, with clients consolidating workflows into Metal and retiring legacy systems.

  9. 24:48 – 26:40

    Data Decay and CRM Maintenance

    Speakers discuss how CRM data naturally decays and requires constant maintenance, likening the process to 'spring cleaning' or adding salt to cooking.

  10. 26:40 – 28:20

    Internal Data Trust and Metal's Headless CRM

    Lowe explains Metal's 'headless' CRM accessed via email and Slack agents, which transcribes calls and maps data to fields to ensure high trustworthiness.

  11. 28:20 – 30:03

    The 'Poisoning' of Databases and Salesforce Critique

    Lowe claims every data entry 'poisons' a database due to immediate decay, arguing Salesforce is hated because user behavior poisons data daily.

  12. 30:03 – 32:03

    External Data and Benchmarking Challenges

    The conversation shifts to external data, noting the challenge of contextualizing it against internal firm data without normalization.

  13. 32:03 – 33:01

    Deal Profiles and Pattern Recognition

    Metal's platform uses 'deal profiles' that refresh in the background to identify patterns, combined with third-party benchmarks for shared context.

  14. 33:01 – 35:13

    AI Strategy and Education

    Lowe identifies educating clients on the importance of the data foundation before building AI workflows as the biggest challenge.

Claims worth verifying

Dakota Marketplace is a comprehensive institutional and intermediate database built by fundraisers for fundraisers that consolidates all channels and asset classes.

Unidentified speaker (speaker_0) ·0:43

How to verify: Visit Dakota.com or review marketing materials for Dakota Marketplace to confirm its stated features and target audience. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.

Taylor Lowe started as an SDR at Copper, which was then called ProsperWorks.

Unidentified speaker (UNKNOWN) ·3:38

How to verify: Check Taylor Lowe’s LinkedIn profile or company press releases from Copper/ProsperWorks regarding early hires. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.

Primary source

Kustomer was sold to Meta.

Unidentified speaker (speaker_2) ·4:18

How to verify: Search SEC filings or press releases from Meta (formerly Facebook) regarding the acquisition of Kustomer. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.

Primary source

Metal was launched out of Y Combinator in early 2023.

Unidentified speaker (speaker_2) ·4:38

How to verify: Check Y Combinator’s company database or Metal’s official press releases for launch date and cohort. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.

Investment firms do not have a ubiquitous source of record or system of record for their data.

Unidentified speaker (speaker_2) ·8:09

How to verify: Review industry reports on private market technology adoption or survey data regarding data fragmentation in asset management. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.

Metal was initially launched as horizontal infrastructure for large language models before focusing on private markets.

Unidentified speaker (speaker_2) ·4:38

How to verify: Review Metal’s historical press releases or blog posts detailing their product evolution from 2026. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.

Metal’s platform uses explicit classification and provenance tracking to handle document types differently, which generic LLMs cannot do out of the box.

Unidentified speaker (speaker_2) ·11:32

How to verify: Review Metal’s technical documentation or whitepapers describing their data architecture and handling of CIMs vs. QoEs. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.

Metal’s data is accessible via MCP in environments like GPT and Gemini.

Unidentified speaker (speaker_2) ·13:10

How to verify: Check Metal’s integration documentation for Model Context Protocol (MCP) support and compatibility with OpenAI and Google models. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.

Metal does not perform training of any sort within its platform.

Unidentified speaker (speaker_2) ·21:46

How to verify: Review Metal’s privacy policy or technical documentation regarding data usage and model training practices. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.

Some customers use Metal as a central node, effectively retiring legacy systems like their primary CRM or file repository for certain workflows.

Unidentified speaker (speaker_2) ·24:17

How to verify: Review customer case studies published by Metal or independent reviews of their implementation outcomes. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.

Salesforce is considered the 'greatest platform in the world' for sophistication and utility.

Unidentified speaker (UNKNOWN) ·28:42

How to verify: Market research reports on CRM market share and feature sets, or Salesforce's own marketing materials. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.

Metal uses a 'headless' CRM accessed via email, Slack, and MCP agents.

Unidentified speaker (speaker_2) ·27:20

How to verify: Metal's product documentation, website, or technical whitepapers describing their API and agent architecture. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.

Data in a database 'poisons' the database and decays from day one.

Unidentified speaker (UNKNOWN) ·26:11

How to verify: Data management industry literature on data freshness, staleness, and the need for data governance frameworks. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.

Creating a benchmark without normalized data requires chasing six to seven sources.

Unidentified speaker (speaker_2) ·30:23

How to verify: Case studies or whitepapers from data aggregation firms detailing the manual effort required for data reconciliation. For recent events, verify 2026-08-10 through 2026-11-10. Use another period only when the speaker explicitly names it.

Do your own research

  • Verify Metal's Y Combinator cohort and funding history via YC's official database or Crunchbase.
  • Review Metal's technical documentation on Model Context Protocol (MCP) integration and data provenance tracking.
  • Assess Salesforce user satisfaction and data quality metrics through Gartner Peer Insights or internal IT audits.
  • Compare Metal's 'headless' CRM data capture accuracy against traditional manual entry methods in pilot environments.
  • Investigate the specific document classification algorithms Metal uses to distinguish between CIMs and Quality of Earnings reports.
  • Check for conflicts of interest if Metal partners with specific LLM providers or data vendors mentioned.
  • Review case studies provided by Metal for evidence of reduced vendor consolidation and workflow efficiency gains.
  • Verify claims about data decay rates in traditional CRMs against academic or industry studies on CRM data quality.
Full transcriptexpand
0:10
Unidentified speaker (speaker_1): Welcome to the Rainmaker Podcast with your host, Geek Costin. The goal of this podcast is to give listeners a unique look into sales strategies from top industry executives. We introduce you to the heads of sales and heads of distribution who will help you understand the inner workings of the success. successful sales organizations from philosophy to
0:29
Unidentified speaker (speaker_1): execution. This podcast is essential for sales professionals seeking wisdom from the best in the field. If you're not familiar with Dakota and their Dakota Rainmaker content, please check out Dakota.com to learn more about their services.
0:43
Unidentified speaker (speaker_0): This episode is brought to you by Dakota Marketplace. Are you tired of constantly jumping between multiple databases and channels to find the right investment opportunities? Introducing Dakota Marketplace, the comprehensive institutional and intermediate database built by fundraisers for fundraisers. With Dakota Marketplace, you'll have access to all channels and asset classes in one place, saving you time and
1:03
Unidentified speaker (speaker_0): streamlining your fundraising process. Say goodbye to the frustration of searching through multiple databases, websites, form ADVs, and say hello to a seamless and efficient fundraising fundraiser experience. Sign up now and see the difference Dakota Marketplace. can make for you. Visit Dakota.com forward slash Dakota Marketplace today. What is up, everybody? It's Geek Costin, Founder 2 of
1:23
Unidentified speaker (speaker_0): Dakota. Welcome to the latest Rainmaker podcast. I am so fortunate to be joined by Taylor Lowe, CEO of Metal. Taylor, welcome.
1:32
Unidentified speaker (speaker_2): Thank you. Thank you so much for having me.
1:33
Unidentified speaker (speaker_0): Taylor Lowe is the CEO and co-founder of Metal, a New York-based AI company built. building the context layer for private capital. Metal puts a firm's deal history, relationships, and judgment into every AI tool it uses. This is a very unique conversation we normally have on distribution leaders. And
1:54
Unidentified speaker (speaker_0): the whole goal is how do we appeal to the one-person firm, right? One person fundraiser at a boutique investment firm. What can we be helping them learn? Best practices, right? However, today this is very, very related but different because now one of like every
2:14
Unidentified speaker (speaker_0): head of distribution
2:15
Unidentified speaker (UNKNOWN): is
2:15
Unidentified speaker (speaker_0): trying to figure out what am I doing with AI, LLMs, my data, external data sources, cleaning it up, all of this. And we see it because these are vexing, vexing problems. Taylor, you've created a tremendous business in metal. And you're solving a lot of these problems. We're going to talk about that today. And
2:35
Unidentified speaker (speaker_0): the reason we wanted to have you come on board is because it's so vital to all the decisions that these distribution leaders are trying to figure out, trying to make. We we watch it in real time. And these are not easy decisions, these are very thorny. You've got so many people weighing in, compliance, all the different stuff, data,
2:55
Unidentified speaker (speaker_0): data integrity, how do you scrub it? How do you bring in external data sources? All of that. But
3:00
Unidentified speaker (speaker_2): yeah,
3:00
Unidentified speaker (speaker_0): before we get to that, right, give us the origin story, right? Where'd you grow up? Give us kind of the career path, college, little thing, family, and then we'll get into metal.
3:08
Unidentified speaker (speaker_2): Yeah, yeah. No, absolutely. And and thanks for that. And yeah, a lot to talk about there, of course. Um for me, yeah, just a bit of background. So I actually, you know, I grew up in Washington DC, uh, so East Coaster. Um, spent some time in San Francisco where sort of learned the ropes in the technology industry and have spent my whole career really um in software,
3:28
Unidentified speaker (speaker_2): enterprise SaaS, mostly on the, you know, data products, infrastructure and API side of things. Um, a lot of that time actually spent in CRM as well. So for
3:38
Unidentified speaker (UNKNOWN): who? Yeah.
3:39
Unidentified speaker (speaker_2): Uh a few actually. So I started um my very first job in San Francisco. I was an SDR at a company called Copper, then it was called ProsperWorks. Um yeah, yeah. Uh I was a great experience there. Um, moved up into an account executive role, uh, mid-market enterprise and all that. Um, and then actually went laterally and joined the product team as a product
3:59
Unidentified speaker (speaker_2): manager. Um from there I then moved to New York. Uh where I joined a company called Customer. Um as a CRM based uh in New York. I was an early employee there, where I also met my co-founders. Uh there for about four or five years, we ended up selling that business to Meta. Um, yeah, really good outcome. But I think, you know, we we learned a ton along the way as well. Very, very
4:18
Unidentified speaker (speaker_2): technically advanced product for for the category. Um and then yeah, from there we actually launched Metal um shortly after the acquisition in the start of 23. We launched Metal out of Y Combinator initially as a just horizontal infrastructure platform for large language models before very quickly becoming much more focused on
4:38
Unidentified speaker (speaker_2): obviously the market that we're talking about today.
4:40
Unidentified speaker (speaker_0): So just for I mean, a lot of people listening to this, especially the tech oriented heads of distribution, will know Y why Combinator. What was it like being part of that?
4:49
Unidentified speaker (speaker_2): Phenomenal. Yeah. I think for me, um, look, a lot of folks in the industry, um, you know, you you sort of get into it from the outside in reading, you know, Paul Graham essays, reading uh YC content. Um, they're just such a luminary within this space, and obviously the the track record speaks for itself. Um, I it was uh just an amazing experience. I think going through uh
5:09
Unidentified speaker (speaker_2): with the wealth of knowledge and leadership within that um group and being able to access. and talk to folks who have
5:16
Unidentified speaker (UNKNOWN): who
5:16
Unidentified speaker (speaker_2): talk with quite literally like hundreds of founders, you know, at any given point.
5:21
Unidentified speaker (speaker_2): They have just seen so many things, right? And they help you ask. I think one of the biggest things is they help you ask the right questions, right?
5:27
Unidentified speaker (UNKNOWN): A
5:27
Unidentified speaker (speaker_2): big part of what you're doing, especially in the early days, there's there's so much doubt. You're trying to figure out like, hey, does anyone even care about this? Right. And it's so easy to kind of knock yourself down. But if you have the right framework and the right set of questions to be honest with yourself and what you're seeing, I think that's one of the hugest um pieces of help that they can offer. And they can very much help steer you. Now you still do the
5:47
Unidentified speaker (speaker_2): work, of course, um as it should be. But um yeah, I think for us, you know, a lot of the frameworks that we learn there, we still use today for sure within metal. Um just a great experience for us. Yeah.
5:58
Unidentified speaker (speaker_0): Any one framework that stands out more than another?
6:01
Unidentified speaker (speaker_2): Yeah, I think um, you know, one thing we we talk about a lot, and we sort of paraphrase it, you know, within our own uh group, which is just look, there's gonna be so much noise, there's always hype, there's some new flashy toy, et cetera, that comes out. Um, our job as founders is to ultimately find and be laser focused on what the truth is in the market, right? Like
6:21
Unidentified speaker (speaker_2): what absent of hype, what absent of whatever's in the news cycle today is is really true that we want to help build this business around. Um and having that sort of um very maniacal focus on, you know, what that truth is and how we can build value around it helps filter out a lot of the noise. If you don't have a framework like that, it can be very, very easy today to get distracted by the next shiny thing.
6:41
Unidentified speaker (speaker_2): Um, but I think that translates into focus, which is one of the superpowers of a good startup.
6:46
Unidentified speaker (speaker_0): Right. So let's fast forward to the customer use case.
6:50
Unidentified speaker (UNKNOWN): Yeah.
6:51
Unidentified speaker (speaker_0): Let's talk about investment firms. And investment firms now, heads of distribution and really CEOs are faced with a massive, massive dilemma. They have whatever it might be, let's just make it simple. Their databases, CRM, they have all this information, not just customer information, not just fundraising information,
7:11
Unidentified speaker (speaker_0): but client information. You might have financial information, right? All this. all information, right? All this data, if you will. And it's their own internal data.
7:21
Unidentified speaker (speaker_0): Walk me through, and I know the question's somewhat obvious, but just so we can get it out in the air, what is the problem with all of that data today?
7:29
Unidentified speaker (speaker_2): Yeah. Yeah, it's a great question. So I think look, you you sort of look at the the evolution of the landscape, right? And the data landscape itself. Um it's always been the hardest thing, but also the most valuable to structure, store, and operationalize. And that is internal data, right? Effectively, like what is unique to each firm, which helps accomplish its
7:49
Unidentified speaker (speaker_2): track record? Well, it's uh the unique judgment, experience, you know, expertise within a space or a subsector, hard fought lessons, you know, wins, losses, everything in between. All of that is stored, you know, a lot of it is in uh unstructured. data in various document types in various systems. Um, you know, the investment world is not one of these cases where you have this
8:09
Unidentified speaker (speaker_2): ubiquitous source of record or system of record that everyone mines from. You have fragmentation and it's very extreme. So you mentioned a few, uh CRM, file repositories, email, a conversation you have with a banker, Or something that was set over Teams, right? Um each and every one of these silos houses something that can be very relevant. but you have to connect it to the other pieces. And you have
8:29
Unidentified speaker (speaker_2): to, in software parlance, you have to normalize that, right? It has to all be sort of speaking the same language from a data perspective. That is a very, very hard problem to solve. It was impossible, frankly, before LLMs and before AI, because none of these systems really talk to each other. They were doing distinct jobs. But LLMs, right, if used correctly in And AI, is that you can actually do that and then build that a really, really solid
8:49
Unidentified speaker (speaker_2): data foundation to then build the next set of workflows and use cases on top of.
8:53
Unidentified speaker (speaker_0): And so what is it that the LLMs do to kind of normalize that data to then make it usable and
8:59
Unidentified speaker (UNKNOWN): useful?
8:59
Unidentified speaker (speaker_2): So I think you know, you got to look at each system a bit of its own case, right? And here's some of the interesting learnings for us. You know, again, we come from 10 plus years in CRM, a lot of our engineering team, CRM backgrounds as well. Um, CRMs provide a ton of structure.
9:15
Unidentified speaker (speaker_2): in terms of how the firm thinks about the world. Forget about like how fields are populated, but you know, what do they classify as like subsectors, themes, focus areas? How do they break down their funds? How do they break down their target lists, right? Those are all really important points of context that if you start to feed in unstructured data from sources like the file
9:35
Unidentified speaker (speaker_2): repository or from email, et cetera, you can actually start to create this unique. picture that tells the firm its own history effectively. So you got to kind of treat those a little bit differently, but from a data architecture standpoint, you've got to make sure that everything's got its right place and that the CRM can provide structure, for example. The file repository can provide a lot of those unstructured insights that can
9:55
Unidentified speaker (speaker_2): be extracted from those documents. But I think the piece on top of that is you can't just do this with a generic LM or AI, right? You have to take very strong opinions. Um, and that substantiates in a lot of different ways throughout the platform. Happy to talk about that, but I can pause there too. I just want to make sure you got
10:11
Unidentified speaker (UNKNOWN): it.
10:11
Unidentified speaker (speaker_0): Right.
10:11
Unidentified speaker (UNKNOWN): But
10:11
Unidentified speaker (speaker_0): let's but keep keep going on
10:13
Unidentified speaker (UNKNOWN): this
10:13
Unidentified speaker (speaker_0): topic because essentially what you're doing is you're taking all this data an organization has,
10:19
Unidentified speaker (speaker_2): yeah,
10:20
Unidentified speaker (speaker_0): and
10:20
Unidentified speaker (UNKNOWN): you're
10:21
Unidentified speaker (speaker_0): you're you're you're effectively normalizing it so then it actually can be, you can put in an another LLM. on top of it. Correct.
10:28
Unidentified speaker (UNKNOWN): So
10:28
Unidentified speaker (speaker_0): you're actually using an LLM to normalize the data so it's actually digestible. It's LLM a bowl, right?
10:35
Unidentified speaker (UNKNOWN): I'm having
10:36
Unidentified speaker (speaker_0): fun with that. And and so but that that that's a huge service because right now no one kind of knows where to go with all this. It's all so crazy.
10:45
Unidentified speaker (UNKNOWN): And
10:45
Unidentified speaker (speaker_0): you you have a lot of hard hardworking working people. These companies just don't have the technical know know-how how. or understanding of how to actually normalize that data, nor
10:54
Unidentified speaker (speaker_2): how
10:55
Unidentified speaker (speaker_0): many leaders want to put the responsibility on an employee versus turning it over to a consultant who could you know potentially do that for you.
11:03
Unidentified speaker (speaker_2): Right.
11:03
Unidentified speaker (UNKNOWN): Right.
11:03
Unidentified speaker (speaker_0): And so but so
11:05
Unidentified speaker (UNKNOWN): so
11:05
Unidentified speaker (speaker_0): the so the value is how can you take all this data that we have and then put it in in a place where it an actually action can be acted on?
11:12
Unidentified speaker (speaker_2): Yeah, exactly. No, and I think you know you mentioned a few pieces there. Like the LM, of course, has a big role, but there is a ton of software that we build around that to make it work, right? So a few of the things that we would do is like classify the types of documents, right? We're gonna want to handle, for example, like financial metrics from a sim versus a Q of E differently. Those things have
11:32
Unidentified speaker (speaker_2): different ranking and provenance within our systems. Um, so classification, how we then extract the right data out of those types of documents as well, all has to take a bit of an opinion. LMs don't come out of the box with that degree of nuance and specificity, let alone how a single firm would want to do it, given their focus area or their strategy or the types of teams they
11:52
Unidentified speaker (speaker_2): have set up. That has to be written explicitly around the software in concert with the LLM.
11:57
Unidentified speaker (speaker_0): Gotcha.
11:58
Unidentified speaker (speaker_2): Yeah.
11:58
Unidentified speaker (speaker_0): And okay, then then let's just now go to and you guys have. decided to take a very different approach around your agnostic.
12:08
Unidentified speaker (speaker_2): Yeah.
12:09
Unidentified speaker (speaker_0): Right to what LLM a company would use. And then, you know, like for instance, there has been situations you're looking at decoded data and you or you could have other data sources you
12:18
Unidentified speaker (UNKNOWN): want
12:18
Unidentified speaker (speaker_0): to bring into that. So once you normalize it and sort of cleanse it, organize it, it And now makes it much easier to bring in external data sources, but you're not forcing an anyone into your own personal LLM.
12:30
Unidentified speaker (speaker_2): Correct. Yeah. No, that's a really good point. So look, at the end of the day, um, we view LLMs as a component of the system. They're absolutely critical, of course, but it is a component. It's not the entirety of the system. So I think, you know, one of the pieces there too is that as we're looking at this in the very early days with Metal, we said, look, you know, these foundation labs are going to be competing. There
12:50
Unidentified speaker (speaker_2): could be a third or a fourth entrant that we don't know of that comes out of the blue at some point, right? We don't want to be married to any one of these providers because who knows which way they're going to go. But at the same time, I think the bigger decision there is we don't want to compete with these model providers. We want to use and benefit from the advancements of the labs, but what do we then supply to that stack? And again, it's that fundamental issue of the
13:10
Unidentified speaker (speaker_2): data itself. So that's why our customers, right, can access all of this container. contextualized data, the workflows you run on top of it, et cetera, in our platform, but predominantly they're accessing it more and more within our cloud experience, our GPT experience, Gemini, all through MCP. It's all hitting the same data, but then it's portable, right? So folks can access it wherever they like.
13:30
Unidentified speaker (speaker_2): Email, Slack, Teams, whatever. It's it's um yeah, you can deploy it in any which way you'd like.
13:35
Unidentified speaker (speaker_0): So
13:36
Unidentified speaker (UNKNOWN): you
13:36
Unidentified speaker (speaker_0): started this in twenty 2023. twenty three.
13:37
Unidentified speaker (speaker_2): Mm-hmm.
13:38
Unidentified speaker (speaker_0): Okay. So obviously very close to one chat.
13:40
Unidentified speaker (speaker_2): Yeah.
13:41
Unidentified speaker (speaker_0): Right. Walk us through the initial and how it's grown, conversations with prospects, then became customers. What were concerns early days as people were trying to figure this out? How has their thinking evolved over time? And what what basically could the audience learn from sort of the step up in sort of
14:01
Unidentified speaker (speaker_0): knowledge and recognition of what people really need to solve for
14:04
Unidentified speaker (speaker_2): to
14:04
Unidentified speaker (speaker_0): be able to take advantage of these tools for the advantage of their business as well as their customers?
14:09
Unidentified speaker (speaker_2): Yeah, it's a great question. So a couple parts to that. Um thinking back to some of the early days, right? So I think one of the very first use cases when we started to work with private markets, we started to work with a you know leading uh mid-market firm. Uh, and one of the very first use cases is that we they wanted to process an analysis. and analyze expert calls faster. Um, so
14:29
Unidentified speaker (speaker_2): we then built a file uploader. We thought we were hot stuff at the time because GPT didn't have one. So we're like, oh, you
14:34
Unidentified speaker (UNKNOWN): know,
14:34
Unidentified speaker (speaker_2): we've cracked the code. Obviously, at this point, that's beyond table stakes. But um, so we started to to funnel these expert calls within Metal at the time. They would interact with them, chat with them, very, very basic stuff today, right? But we were kind of exploring and going down the rabbit hole a bit with them. And they're saying, look, you know,
14:49
Unidentified speaker (UNKNOWN): this
14:49
Unidentified speaker (speaker_2): is gonna going save to us X, Y, and Z time. but the bigger thing that we're after is not time, it's insights. It's how we can actually use this information to have better judgment, better investment decisions throughout the process. And that's one of the through lines, right? I think
15:00
Unidentified speaker (UNKNOWN): that's a good thing. Oh,
15:00
Unidentified speaker (speaker_0): so you started expert networks, expertise.
15:04
Unidentified speaker (speaker_2): Just as a document type.
15:05
Unidentified speaker (UNKNOWN): Yeah, no, I got you. But that
15:06
Unidentified speaker (speaker_0): was to help the actual investment side
15:09
Unidentified speaker (UNKNOWN): of the Or house
15:09
Unidentified speaker (speaker_0): make better investment decisions with greater intelligence and insight.
15:12
Unidentified speaker (speaker_2): Yeah, correct. Yeah. And then so we did start a little bit more on like the top of funnel and diligence side, um, to be to be clear. Um, but at this point, we've expanded the platform and are wall to wall with firms, you know, from BD to IR, of course, and then everything in between.
15:27
Unidentified speaker (UNKNOWN): We
15:27
Unidentified speaker (speaker_0): don't
15:27
Unidentified speaker (UNKNOWN): just
15:28
Unidentified speaker (speaker_0): I want to stay on this point for one second because you're doing a uh extraction of you know the the um expert network uh commentary, helping you know, just extract insights.
15:38
Unidentified speaker (UNKNOWN): It
15:38
Unidentified speaker (speaker_0): seems to me, right, what was valuable, having fun with the term, added value six months ago, a year ago, becomes so extinct so quickly and could become so table stakes so fast.
15:49
Unidentified speaker (UNKNOWN): Yeah.
15:50
Unidentified speaker (speaker_2): I
15:50
Unidentified speaker (UNKNOWN): mean,
15:50
Unidentified speaker (speaker_0): maybe advancement of the iPhone with you know, what what did it bring together? 20 different things from GPS and maps and weather and all into one place. How like where like it that's moved at a breakneck pace.
16:02
Unidentified speaker (speaker_2): Mm-hmm.
16:03
Unidentified speaker (speaker_0): And what was an advancement six, eight, twelve months ago, is no is now it's just parf the course.
16:10
Unidentified speaker (speaker_2): Right.
16:10
Unidentified speaker (speaker_0): How how do you see everything kind of like how do you interpret the path
16:14
Unidentified speaker (UNKNOWN): of
16:14
Unidentified speaker (speaker_0): this innovation?
16:15
Unidentified speaker (speaker_2): Yeah, no, it's great. So I think it it calls out a a really important distinction, right? And I think and it's also part of your other question on like where is the market today and how are we actually attacking the the AI strategy, as it were? So I think one of the things that folks can get really preoccupied with is at the workflow layer, right? And let's just say a firm came to us today and wanted to do that same thing with an
16:35
Unidentified speaker (speaker_2): expert call and they wanted to mine data from it, et cetera. Um, yeah, you can definitely start and kind of anchor the initial implementation around that workflow. But what you first want to do is actually look at the data foundation, right? Like, where does this information come from? Where is it stored? How does a the firm access it today, right? Because ultimately, you want to pipe that into this data layer next to the other data types, next to the other
16:55
Unidentified speaker (speaker_2): functions, right? Between the investment stat portfolio and IR. That's really the goal, is you want to create this shared context. And that is a data and an infrastructure problem, not a workflow problem. So when folks start with workflows, right, and they don't examine the data layer first. This is the classic case where you know you'll get really excited, you'll get like the 50% fifty percent demo version out there. You say, Hey, look, this
17:15
Unidentified speaker (speaker_2): is great, this is going pretty easily. As soon as you want to start building on that workflow, as soon as you want to start adding in another source that hasn't been included, like you know, CRM, file repo, all these things we've talked about, that's where the complexity really ratchets up and where that data problem becomes very, very front and center. Because these two systems don't speak the same data language. You need pipes effectively to normalize that to
17:35
Unidentified speaker (speaker_2): make them accessible. Um, so that's one of the the tricks, right? All
17:38
Unidentified speaker (UNKNOWN): right, so
17:38
Unidentified speaker (speaker_0): I
17:38
Unidentified speaker (UNKNOWN): want
17:38
Unidentified speaker (speaker_0): to just let's just back up for one
17:40
Unidentified speaker (UNKNOWN): sec
17:40
Unidentified speaker (speaker_0): because back in the day when I would do PowerPoint presentations, right? A lot
17:44
Unidentified speaker (UNKNOWN): of
17:45
Unidentified speaker (speaker_0): people would take the PowerPoint
17:46
Unidentified speaker (UNKNOWN): and
17:47
Unidentified speaker (speaker_0): then would start with the PowerPoint and they would start building from the PowerPoint. Whereas
17:52
Unidentified speaker (UNKNOWN): the
17:52
Unidentified speaker (speaker_0): way that I did it is I would start with a yellow pad,
17:54
Unidentified speaker (UNKNOWN): slide
17:55
Unidentified speaker (speaker_0): one, what point one would make, slide two, what point one do I would want to make,
17:58
Unidentified speaker (UNKNOWN): get
17:58
Unidentified speaker (speaker_0): all the way down to slide 10 or something, 15,
18:01
Unidentified speaker (UNKNOWN): and
18:01
Unidentified speaker (speaker_0): then I would go to
18:02
Unidentified speaker (UNKNOWN): the
18:02
Unidentified speaker (speaker_0): printer. bring out 15 pages, staple them, and then start writing down on each one. And then once that's all done, then you'd actually move to the technological implementation,
18:13
Unidentified speaker (UNKNOWN): right? If you
18:13
Unidentified speaker (speaker_0): will. So I love what you're saying about because I I I'd meet people to do CRMs, right? They would be freaking out about the fields and the UI and the whole thing.
18:23
Unidentified speaker (UNKNOWN): Yeah.
18:23
Unidentified speaker (speaker_2): And
18:23
Unidentified speaker (speaker_0): they didn't care, never even thought about the actual. data. At the end of the day, the design of the UI and the fields are all based upon the data that you have.
18:32
Unidentified speaker (speaker_2): Yep.
18:33
Unidentified speaker (speaker_0): So you're saying some even today, people will jump the gun on the data and go effectively to the design.
18:40
Unidentified speaker (speaker_2): Yeah. I uh of the workflow itself. And I say, look, I uh
18:42
Unidentified speaker (UNKNOWN): I mean
18:43
Unidentified speaker (speaker_0): I'm I'm saying design is workflow.
18:44
Unidentified speaker (speaker_2): Yes, yeah, absolutely
18:45
Unidentified speaker (UNKNOWN): no. I
18:45
Unidentified speaker (speaker_2): totally agree. I think um look, I I take a step back and it's totally logical, it makes sense to
18:50
Unidentified speaker (UNKNOWN): me.
18:50
Unidentified speaker (speaker_2): If I'm if in in that position, right, um, I'd I'd probably do something similar, just having if that was my my background, right? But again, this is more of an unfair comparison because it is a software problem, right? Like you do need to think about these things from a data perspective. But these are unsexy, non-glamorous concepts, right? Like thinking about the plumbing and how we want to classify and you know,
19:10
Unidentified speaker (speaker_2): find provenance and and um resolve certain data conflicts, like all this stuff. It can it can get pretty hairy pretty quickly, right? Um, but it is the thing that actually feeds into the rest of it. So you do need to solve that piece first. Um but I understand why people because you want the the the workflow to work.
19:27
Unidentified speaker (UNKNOWN): Yeah.
19:27
Unidentified speaker (speaker_0): Do you solve that problem for your customers?
19:30
Unidentified speaker (speaker_2): We do.
19:30
Unidentified speaker (UNKNOWN): So
19:30
Unidentified speaker (speaker_0): so you're but also you you have a distinct advantage over of your customers is you're coming in clean.
19:35
Unidentified speaker (speaker_2): Yeah.
19:36
Unidentified speaker (UNKNOWN): And
19:36
Unidentified speaker (speaker_0): there's no agita of this, you just see it cleanly. We should do this.
19:39
Unidentified speaker (UNKNOWN): And
19:39
Unidentified speaker (speaker_0): then everyone on the other side, as you know, right?
19:43
Unidentified speaker (speaker_2): Yeah. Well I think we've got a lot of experience with it too. I mean, you know, every firm is unique, but a lot of the problems are the same. You know, I I've mentioned a few, right? But like the CRM provides a lot of structure and context to how the firm thinks, but it usually is,
19:56
Unidentified speaker (UNKNOWN): you
19:56
Unidentified speaker (speaker_2): know, lacking some data or data fidelity issues because the constant problem with CRMs has always been data entry, right? File repository repositories on the meanwhile, right? Not a lot of structure, just files and folders, but within those files, tremendous context, right? So unifying those two things, you do see a lot of patterns. Um getting that suit up as quickly as we can for customers is a big deal
20:16
Unidentified speaker (speaker_2): for us because that's kind of the first tier of unlocking value on the platform.
20:19
Unidentified speaker (speaker_0): Gotcha.
20:20
Unidentified speaker (speaker_2): Yeah.
20:20
Unidentified speaker (speaker_0): And so you end up obviously can look at it a little more cleanly
20:23
Unidentified speaker (speaker_2): because
20:24
Unidentified speaker (speaker_0): you're not in the muck as much as your customers
20:26
Unidentified speaker (UNKNOWN): are.
20:26
Unidentified speaker (speaker_0): And then how do you get to the eventual reality of where the data really matters to
20:31
Unidentified speaker (UNKNOWN): get
20:31
Unidentified speaker (speaker_0): the data? And and if somebody's listening right now and they're trying to think through all these issues, the promise that you guys have, I don't want to overextend you, but the promise that you guys have to your customers is what in terms of the data?
20:45
Unidentified speaker (speaker_2): Well, I think for one, at the foundational level, we get it all in one place and then it's cleaned and trustworthy, right? Out of the gate. So I can't tell you how many times when we start with a customer, we'll start pulling in this data and they'll
20:58
Unidentified speaker (UNKNOWN): say,
20:58
Unidentified speaker (speaker_2): Hey, look, you know, where'd you get this? I think this was a hallucination. And we'll say, it actually came from your source system. And then we'll look at that back and forth and realize that okay, this is actually an input capture problem, right? Um, and that's something that our platform can help with. But it also helps open their eyes up to this is the data that we've been working with. for years, right? And it actually exposes, I think, in a more
21:17
Unidentified speaker (speaker_2): granular detail how much trust issues or how many trust issues there are within that. Um, which again, LM's and with our platform can do is um quite helpful in rectifying that.
21:27
Unidentified speaker (speaker_0): So again, from the listener's perspective and my personal perspective, once they start working with your platform, do you see a constant progression upwards?
21:37
Unidentified speaker (speaker_2): Yeah. Meaning
21:37
Unidentified speaker (speaker_0): it's not it's not a situation. where you're going backwards, you're going up and you're taking a step back. You're going up, it's just a constant progression because it's learning, it's growing, it's learning, it's growing.
21:46
Unidentified speaker (speaker_2): Yeah, it's it's one of those effects where, you know, the more data, the more context that we provide it, the more it has access to and the more it can effectively learn from that. Well, and I I say learning carefully because there's no training of any sort happening within our platform. Um, but you know, the the context can connect the dots between itself and then it'll make the next deal that you know you
22:05
Unidentified speaker (speaker_2): passed on, the next LP that you passed on or passed on you rather, you're you're smarter for that. So there is that compounding flywheel effect. As folks are actually leveraging it, whether it's uh chat search, running workflows, building dashboards around their their pipeline or fundraising process, whatever it might be, um, that then feeds more data into the platform, right? Where it actually becomes then
22:25
Unidentified speaker (speaker_2): sort of that that sense central node uh within the firm's intelligence.
22:29
Unidentified speaker (UNKNOWN): Yeah,
22:29
Unidentified speaker (speaker_0): so so I'm I'm a golfer. Yeah. Are you?
22:32
Unidentified speaker (speaker_2): Um some would call it that, but it's
22:34
Unidentified speaker (UNKNOWN): quite
22:34
Unidentified speaker (speaker_2): poor. Yeah.
22:35
Unidentified speaker (speaker_0): So for the past 25 years, right? Um, and some people might not like for me to say this, but this is just facts and reality.
22:43
Unidentified speaker (UNKNOWN): And
22:43
Unidentified speaker (speaker_0): it's going a lot, a lot of around growing grass. But this this goes exactly to how I'm listening to you, is that these clubs have had to take on the trees.
22:54
Unidentified speaker (speaker_2): Yeah.
22:54
Unidentified speaker (speaker_0): Right. Yeah. From airflow, dying grass, sunlight, there's all these things, right? A lot of people aren't happy when trees are torn down, right?
23:03
Unidentified speaker (speaker_2): Yeah.
23:03
Unidentified speaker (speaker_0): But but the reality is just for the health of the the the turf, the maintenance, the the water, the irrigation, all of that need they so over time they end up taking trees down. But it's not generally all at once. It's not all in one year, but it's
23:17
Unidentified speaker (UNKNOWN): a
23:18
Unidentified speaker (speaker_0): progression.
23:18
Unidentified speaker (speaker_2): Right.
23:19
Unidentified speaker (speaker_0): Right. And
23:19
Unidentified speaker (UNKNOWN): they they
23:19
Unidentified speaker (speaker_0): have a plan in place and it's a progression and they try to do it thoughtfully and the whole thing. And then, but what ends up happening is over time, all you do is you see an increase in the quality
23:30
Unidentified speaker (speaker_3): of
23:30
Unidentified speaker (UNKNOWN): the
23:30
Unidentified speaker (speaker_0): grass, the turf, the irrigation. There's so many benefits to the environment, right? When you get airflow coming in. My point is, I know it's a simple, simple example. but as I'm listening to you about uh sort of the progression of someone's data, essentially working with your platform, it's almost like every single day progress is being made
23:50
Unidentified speaker (speaker_0): towards an ultimate goal of having be trustworthy.
23:53
Unidentified speaker (speaker_2): Right.
23:53
Unidentified speaker (speaker_0): So then it can get way better answers out of whatever whatever LLM they choose.
23:57
Unidentified speaker (speaker_2): Yeah, 100%. And I think, you know, so establishing that trust, you know, through all the infrastructure pieces we just talked about, very critical. But then keeping in mind, like what is the ultimate goal? And it's one of the things I mentioned earlier, which is, you know, these firms run on judgment, right? At the end of the day, we're trying to get to judgment faster based on our own conviction, based on our own experience. And you have to actually make
24:17
Unidentified speaker (speaker_2): that programmable, right, within the firm. So, you know, as as you're bringing that up and talking about like removing trees, we've had cases where customers will look at some of the source systems. that are feeding metal and saying, look, we're just basically using this thing as like bucket one goes to metal, and then we're writing data back to bucket one, but it's all just being accessed and worked with in metal. Why do we
24:37
Unidentified speaker (speaker_2): still have bucket one? Right. So you actually can shrink and reduce the surface area of the services you're using, the products you're using, um, dependent on the use case, right? Um, I think that kind of fits with your analogy.
24:48
Unidentified speaker (speaker_0): Yeah.
24:48
Unidentified speaker (speaker_2): Um, and then yeah, and then so as that canonical data. layer, right? That's some of the value we can provide uh just to simplify that for them.
24:55
Unidentified speaker (UNKNOWN): Yeah,
24:55
Unidentified speaker (speaker_0): because I mean I think of it in our upstairs office, we always tease we do, you know, spring cleaning.
25:00
Unidentified speaker (UNKNOWN): Right.
25:00
Unidentified speaker (speaker_0): Now, in some ways I say, you know, we shouldn't have to do spring cleaning because let's just not create the problem
25:06
Unidentified speaker (UNKNOWN): in the first place.
25:07
Unidentified speaker (speaker_0): But most CRMs always need a cleansing, right? And and and it seems to me, and what what I'm trying to make the case for for what you're doing. is that this is this is a progression up and to the right, which is a very different experience than anyone has had as relates to CRM in the past.
25:23
Unidentified speaker (UNKNOWN): Yes.
25:24
Unidentified speaker (speaker_2): Because
25:24
Unidentified speaker (speaker_0): if CRM has we're gonna do a blitz, we're gonna do this, and
25:28
Unidentified speaker (UNKNOWN): then
25:28
Unidentified speaker (speaker_0): boom, we're back off to our jobs. Yep.
25:30
Unidentified speaker (speaker_2): Right.
25:30
Unidentified speaker (UNKNOWN): And what
25:31
Unidentified speaker (speaker_0): ends up happening? Decay.
25:32
Unidentified speaker (speaker_2): Yeah. Decay,
25:33
Unidentified speaker (UNKNOWN): decay, decay. Right. It's constant maintenance. And I
25:35
Unidentified speaker (speaker_0): think I think that's what people underestimate about data. Because from where I sit, people ask me all the time, how do you how do you get the data?
25:43
Unidentified speaker (speaker_2): Yeah. And
25:43
Unidentified speaker (speaker_0): I say, look, been doing this seven years. I have to be a little bit rude in my answer. Because I'm like, you're asking the wrong question.
25:50
Unidentified speaker (speaker_2): Yeah.
25:50
Unidentified speaker (speaker_0): Right. Data acquisition is not the hard part, right? Because data is abundant, data is publicly available. There's all sources of how you can actually get data.
25:59
Unidentified speaker (UNKNOWN): Um,
25:59
Unidentified speaker (speaker_0): There's two other components, and the third component's the most important. The second component is you need to be able to bring the data in, you need to be able to enrich it and organize it.
26:07
Unidentified speaker (UNKNOWN): Yeah.
26:07
Unidentified speaker (speaker_2): Right.
26:07
Unidentified speaker (speaker_0): So it can actually make sense. And now in the world of an LLM,
26:11
Unidentified speaker (UNKNOWN): it's
26:11
Unidentified speaker (speaker_0): even more important. But really, at the end of the day, every time you bring in a piece of data into your database, you poison your database. Right. And why is that? Because it starts to decay day one.
26:23
Unidentified speaker (speaker_2): Yeah.
26:23
Unidentified speaker (speaker_0): It's poisoned. poison. Now
26:24
Unidentified speaker (UNKNOWN): you
26:25
Unidentified speaker (speaker_0): can have an antidote, which is a very robust updating system.
26:28
Unidentified speaker (speaker_2): Yep. Right.
26:29
Unidentified speaker (UNKNOWN): And
26:29
Unidentified speaker (speaker_0): and that can be the antidote to the poison. Right. But make no mistake about it. Don't brag about how much data you have. Because all you're
26:36
Unidentified speaker (UNKNOWN): really
26:36
Unidentified speaker (speaker_0): saying is you're raising your hand and saying, I have an updating problem.
26:38
Unidentified speaker (speaker_2): Yeah. Yeah, exactly. Exactly.
26:40
Unidentified speaker (UNKNOWN): I
26:40
Unidentified speaker (speaker_2): mean, look, I liken it often to if if you're doing the manual version of this, it's like salt and cooking. Easy to add, hard to pull out. So you got gotta to be really diligent about that. Unless, to your point, you have a system which actually triages it, maintains it, right, and removes that input um issue. Right. So I mean, maybe one thing I can talk about even is, you know, again, 10 plus years in CRM for me.
27:00
Unidentified speaker (speaker_2): I started my career as an SDR. It's the same pitch within CRM, which is like, hey, uh, you know, we're gonna have unparalleled visibility into your business. Here's gonna going be to all the reports we can run, the pipeline visibility, how we can enact on that, et cetera. The problem is, is that if everyone puts in the right data, which is the big if, which never really quite takes place, I can
27:20
Unidentified speaker (speaker_2): tell you right now our own use of our own CRM at Metal. Um, we've removed those problems. And I use the CRM, it's completely headless. I access it within many different services. I have, I cannot tell you the last time I logged directly into the CRM. I'm accessing it through our email agents, through our Slack agents, um, through our uh MCP. wherever it might be, or
27:40
Unidentified speaker (speaker_2): wherever I am. Um, but I know I can trust that data because we have so many agents as well running through our business, which capture the right inputs from transcribing calls to pulling out the key points, um, to mapping to the right fields in our CRM. So that data is very trustworthy. And then I can operate with it in the same way that I can start to pull in data outside of the CRM to have a really, really
27:59
Unidentified speaker (speaker_2): broad view, 360 view of the business. I think that's kind of a microcosm of what we are doing with our customers. It just so happens that they are dealing to the tune of millions and millions and millions of files, right? Where we as a startup, right, uh don't particularly have that type of volume. So yeah.
28:14
Unidentified speaker (speaker_0): But but but getting back to the point of starting with a metal day one.
28:20
Unidentified speaker (UNKNOWN): Yeah.
28:20
Unidentified speaker (speaker_0): And every day, what we're what we're getting at is getting better, and we're not in the decay business because it keeps moving up into the and to right. And that's
28:27
Unidentified speaker (UNKNOWN): why
28:27
Unidentified speaker (speaker_0): it's just making the case is that everyone loves a good CRM project and it's then it's dead on arrival because
28:33
Unidentified speaker (UNKNOWN): it
28:33
Unidentified speaker (speaker_0): decays the more the sales people enter data, the more it goes out of date, the more other inaccurate it might be. And that's why everyone hates Salesforce, which
28:42
Unidentified speaker (UNKNOWN): is
28:42
Unidentified speaker (speaker_0): bit basically which I laugh about because it's like the greatest platform in the world. If you look at it for right, it's sophistication utility.
28:49
Unidentified speaker (UNKNOWN): Yep.
28:50
Unidentified speaker (speaker_0): But from a usability standpoint, the bait the way the people use it, they poison it every day. And everyone blames Salesforce. And you really gotta be blaming the salespeople or the people entering the information.
28:59
Unidentified speaker (speaker_2): Yeah.
29:00
Unidentified speaker (UNKNOWN): And
29:00
Unidentified speaker (speaker_0): and then the infrastructure of how it's designed, right?
29:03
Unidentified speaker (speaker_2): Right.
29:03
Unidentified speaker (speaker_0): Because we've had you know a magical relationship with Salesforce, both from a user perspective as well as a um uh a a product perspective, if you will. And but we use it properly, right?
29:17
Unidentified speaker (UNKNOWN): And we force
29:17
Unidentified speaker (speaker_0): ourselves to use it properly. And not everyone does that. But that that being said, talk to us about external data sources. So you get the the core
29:29
Unidentified speaker (UNKNOWN): of
29:29
Unidentified speaker (speaker_0): trustworthy, you get it to where you want, right? Forget us for for for one second. What are now the capabilities that you didn't that the your client didn't have? before?
29:39
Unidentified speaker (speaker_2): Yeah.
29:40
Unidentified speaker (speaker_0): Right before the data was trustworthy and organized. Now it is, and you're maintaining
29:45
Unidentified speaker (UNKNOWN): it.
29:45
Unidentified speaker (speaker_0): Talk about kind of the now it's fun. Now,
29:48
Unidentified speaker (UNKNOWN): now we're
29:48
Unidentified speaker (speaker_0): in the world of fun because now if you're connecting external data sources, you're assuming the people delivering those external data sources will bring you clean, up-to-date
29:57
Unidentified speaker (UNKNOWN): Up
29:57
Unidentified speaker (speaker_0): to date data on a consistent basis that now can massively enrich
30:01
Unidentified speaker (speaker_2): the
30:01
Unidentified speaker (speaker_0): data set that you've created for. your customers.
30:03
Unidentified speaker (speaker_2): Yeah, absolutely. And look, and there's obviously tremendous value to the external data market, right? I mean, it's more than ubiquitous, right? Folks are relying on this stuff every day. I think that the missing half of the coin, right, has been how do we contextualize what we're seeing in the broader market or through benchmarks, right, of third-party sources and what we have seen as a firm. Right. And I think one of the challenges, right, when you have this data
30:23
Unidentified speaker (speaker_2): fragmentation problem, you haven't solved it, um, you're trying to create a benchmark, right? Where you then have to chase down six, seven sources. You got to go talk to the the deal team that worked on that, you gotta go talk to CFO, whatever it might be. That might take you a week, right? And then you got two more days for reconciliation. Who knows? Um, you know, that that's all time that you're spending, right? Whereas if you're actually like starting to
30:43
Unidentified speaker (speaker_2): normalize this data into one place, um, that's the single place that you're going to work on top of, right? So I think it's just kind of like a a different framing. of how you can actually look at getting data into your business and focusing on that input layer itself will pay itself back then some. Yeah.
30:59
Unidentified speaker (speaker_0): Yeah. And I think kind of the big takeaway, outside of just you guys being very sophisticated, is that there's actually a relatively simple path forward
31:09
Unidentified speaker (UNKNOWN): to
31:09
Unidentified speaker (speaker_0): take your existing whole conglomeration of data information and in whatever databases that it's in. And you've created a clear path forward to get that normalized. It might take some time. Cause that that I think is like the hope thing. Like, is there is there really hope here? And how many consultants,
31:29
Unidentified speaker (speaker_0): and I love consultants, but you take people down a bad road. They don't really have
31:34
Unidentified speaker (UNKNOWN): an
31:34
Unidentified speaker (speaker_0): outcome to actually get you to where you want to be. Because Cause if you think about it,
31:38
Unidentified speaker (UNKNOWN): if
31:39
Unidentified speaker (speaker_0): you can get someone's data to a place where you can add these external data sources and then use the LLM on that.
31:45
Unidentified speaker (UNKNOWN): I
31:45
Unidentified speaker (speaker_0): mean, how about just use the LM on the clean data first?
31:47
Unidentified speaker (speaker_2): Yep. Yep.
31:48
Unidentified speaker (speaker_0): And
31:48
Unidentified speaker (UNKNOWN): then
31:48
Unidentified speaker (speaker_0): two, bring in one or two or three external data sources that are meaningful to you.
31:52
Unidentified speaker (UNKNOWN): Yeah.
31:53
Unidentified speaker (speaker_0): Now you can start to change the game.
31:55
Unidentified speaker (speaker_2): So I I that's a really good point. I think, you know, one of the capabilities, for example, on our platform that helps substantiate that is we have this concept of a deal profile, right? That is sort of constantly refreshing in the background. Um, the more like deals run through, and this applies to IR as well. We we apply this effectively like pattern recognition that starts to kind of change the more data we put
32:14
Unidentified speaker (speaker_2): into this, uh, helps them understand, like, here are the types of deals, right, or LPs, et cetera, that you've worked well with um or not. Um, and then you can supplement that as well with third-party third party data, right? So you can take the external. benchmarks and and how folks are actually stacking up. You can take your internal performance and what the patterns they're in uh look like. And then that's sort of that shared
32:34
Unidentified speaker (speaker_2): context that you can start to build off of. So I think that the third party data piece is hugely additive, especially when you combine it with the unique insight that each firm has. And
32:44
Unidentified speaker (speaker_0): but it's clean. But you guys
32:45
Unidentified speaker (UNKNOWN): Yeah, exactly. are clean exactly yeah.
32:46
Unidentified speaker (speaker_2): And everything is traceable. You can you can audit it of course. Um yeah that that trust uh um aspect is is very important.
32:52
Unidentified speaker (speaker_0): Okay, last question I always ask everyone. Okay.
32:55
Unidentified speaker (UNKNOWN): Uh
32:55
Unidentified speaker (speaker_0): and it's a simple one, but
32:57
Unidentified speaker (UNKNOWN): uh
32:57
Unidentified speaker (speaker_0): biggest challenge you're facing today and how you're overcoming it within the business.
33:01
Unidentified speaker (speaker_2): Yeah, it's a great question. Look, I think um for us, I mean, we you know, we are extremely confident we have the best product in the market. Um I think for us, you know, making sure that folks can understand there there's a bit of an education piece, right? And in the data problem, the infrastructure problem, right? Like I said, can get very boring very quickly. You don't want to put people to
33:21
Unidentified speaker (speaker_2): sleep on calls, but you've got to also thread the line and say, look, this is why it's important. This is why when we think about spinning up AI initiatives as projects, not products, we can run into walls because we have to take this data foundation view first before we really want to dive into workflows. I think that's that's a big piece of it, right? And working with folks. And, you know, we get asked asked all the time
33:41
Unidentified speaker (speaker_2): in terms of, hey, what's our like greatest competition? Par for the course, nine out of 10 calls, even more than that, is still just building with the foundational labs, trying to enact workflow AB and C and they're actually running into walls. That's that's the status quo today. So helping them understand uh what they can build on top of it with this shared infrastructure. Um that's one of the main things we're focused
34:01
Unidentified speaker (speaker_2): on.
34:02
Unidentified speaker (speaker_0): Well
34:02
Unidentified speaker (UNKNOWN): a
34:02
Unidentified speaker (speaker_0): key takeaway from me in listening is that you really have, you guys have created a capability to solve a really, really big problem for organizations in a very trustworthy way, meaning a way that it's not just, hey, can these guys get it done? You've proven time and time again. Yeah.
34:17
Unidentified speaker (speaker_2): If
34:17
Unidentified speaker (UNKNOWN): you're
34:17
Unidentified speaker (speaker_0): customers, you can see them to the end. So congratulations on what you built. Thanks for the insights and loved having you on the podcast.
34:22
Unidentified speaker (speaker_2): Appreciate it. Thank
34:23
Unidentified speaker (UNKNOWN): you so
34:23
Unidentified speaker (speaker_2): much. Amazing.
34:24
Unidentified speaker (speaker_0): All Alright, right, everyone. That's a wrap of the latest episode. of the Rainmaker Podcast. Thanks so much for joining. We'll see you on the next episode of the Rainmaker Podcast.
34:33
Unidentified speaker (speaker_1): You can find this episode and others on Spotify, Apple, or your favorite podcast platform. We are also available on YouTube if you prefer to watch while you listen. If you would like to check up on past episodes, check out our website, Dakota.com. Finally, if you like what you're hearing and seeing, please be sure to like, follow, and share
34:53
Unidentified speaker (speaker_1): these episodes. We welcome all your feedback as well. Thank you for investing your time with Dakota.
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.

More from All Things Fundraising

All Things Fundraising

It’s Not the Answers, It’s the Questions: Sales Lessons from Bill Wolfe

Episode discussion: Bill Wolfe and Gui Costin discuss sales strategies, emphasizing personal responsibility, preparation, and question-driven due diligence.

Speakers’ claims: Bill Wolfe and Gui Costin share insights on capital raising, identifying personal responsibility as the primary trait of successful salespeople. They emphasize the importance of thorough preparation, advising against asking prospects to explain their business. The speakers highlight the use of CRM systems to measure activity and the value of authentic personal connections in building trust. They also discuss the power of referrals and the strategic use of open-ended questions to drive conversations during due diligence meetings.

Sales StrategyPersonal ResponsibilityCRM Usage
34 min0 assets16 claims to verify
All Things Fundraising

UK Charity Market Webinar

Episode discussion: Dakota expands UK charity and European subadvisory datasets to map $234B in charity AUM and trillion-euro subadvisory market

Speakers’ claims: Rob introduces James Goodman, who details Dakota's expansion into the UK charity market, adding 1,600 accounts with over $234 billion in AUM, 91% of which are new to the platform. Goodman explains that the data is sourced from real-time audited annual accounts, allowing for immediate tracking of trustee decisions and manager changes. The episode also covers the European subadvisory market, estimating one trillion euros managed through platforms in Ireland and Luxembourg, with new data mapping 139 platforms and over 3,000 decision-makers. Future roadmap items include Solvency II insurers, UK private pensions, and family offices, noting the UK has overtaken the US in family office numbers.

UK Charity MarketEuropean SubadvisoryFundraising Data
17 min0 assets10 claims to verify
All insights