Launch Nephilia: Select investors via MCP directly within your own AI assistant

Source: Startup Insider – Original Episode (episode in German)

This time, the focus is on a project of their own: Björn Rieckhoff introduces Nephilia, an information service for founders that helps them prepare for a funding round—specifically, by identifying which investors are actually a good fit for that round. The product deliberately has no interface of its own: Nephilia is an MCP server that is integrated as an information layer into the AI assistants that founders use anyway. In conversation with Jan Thomas, Björn explains why the list of investors will still come to mind as an email with five names even in 2026, what distinguishes a “qualified” shortlist from a keyword search, and why Nephilia provides factual data from the commercial register and deployment behavior, but not investor ratings in the style of Kununu. Also covered: the limitations of historical data when it comes to new sectors like defense tech, the deliberate decision against an “agentic fundraising framework” that handles outreach, and the question of whether seed strapping is shrinking the market for a VC tool. Finally: pricing as a flat fee per funding round instead of a subscription—900 euros for Pre-Seed up to 2,500 euros for Series A—and the realization that building the MCP server itself was the smallest part of the work.

Key themes in this episode

Transcript

This transcript has been edited for readability. The content and statements have not been changed. The original conversation was in German; this is an English translation.

Jan Thomas: Björn, hi, nice to meet you.

Björn Rieckhoff: Hi Jan, nice to meet you too.

Jan Thomas: I have to say, it’s great to have you here. And it’s great what you’ve brought to the table—a really exciting topic. I was just thinking about it again: Whenever I’ve introduced you on this podcast over the past few months, we’ve always mentioned that you were originally with Cavalry Ventures and are now a business angel and advisor. That’s changing right now.

Björn Rieckhoff: A little bit, I feel. That’s changing a bit right now. I’m moving into more of an operational role—we talked about this in our preliminary conversation, as so many current or former VCs are doing right now. I’m actually in the process of setting up an information service. It grew out of my consulting work, because I kept seeing the same problem there—we can discuss exactly what that is in detail in a moment. And I’m currently building nephilia.ai as an information service for founders to support their fundraising efforts.

Jan Thomas: I always tell people who want to advertise on a podcast: You have to pay attention to pronunciation. Nephilia isn’t spelled with an “F,” but with “PH.” The link is in the show notes, and you can sign up there right now—I just wanted to mention that ahead of time.

Björn Rieckhoff: It’s basically a public beta right now, if you will—as we speak.

Jan Thomas: Exactly. As I mentioned, the link is in the show notes, so let’s get started. This is really interesting. You just said it grew out of a problem. I think it’s a good idea to start with the problem.

Björn Rieckhoff: I actually came across a great quote from Florian Heinemann beforehand. He likes to ask, “What is my right to play?”—as a way to make a decision. And I thought that was quite fitting, because I’ve asked myself: Why am I actually building something myself? I have the impression that with Nephilia, a lot of things that have accumulated over the last ten years are coming together for me. As you already mentioned, I was the first employee at Cavalry back then; I helped build the fund, made many investments, and oversaw many transactions—all in all, I believe that’s now close to 100 transactions if you include my consulting work. And in my advisory work, I often asked myself: How can we actually make this more widely accessible, help more founders build on this experiential knowledge? The result is a product that supports founders in the operational preparation of their funding rounds. Nephilia is essentially an MCP that’s integrated as an information layer into a founder’s regular AI assistant. The founder uses their Claude, ChatGPT, or whatever they use—let’s just stick with OpenAI for now—and integrates a plugin—essentially Nephilia—which provides a layer of information within their normal interface. This layer gives them the necessary details about the investor landscape out there so they can decide which investor is the right fit for their round at that moment.

Jan Thomas: I have Nephilia open right now, and at first glance, it looks like you could use it without MCP. But the MCP feature—I already told you this when we spoke on the phone—I have to say, I think it’s very clever. I think it fits perfectly with people’s habits: When I think about our team or many others I know, they basically never leave Claude, ChatGPT, or Claude Code anymore. That means you really want to make sure you pull the information you need right there.

Björn Rieckhoff: Exactly. I think it’s become obsolete by now—and I’m not just talking about Nephilia here, but about startups in general: There’s a whole segment of startups where I simply don’t see any reason for an interface anymore. You no longer need a traditional SaaS interface; you don’t have to struggle at all to build or copy one—it’s more about making information available. And in that regard, MCP as a protocol is essentially the best way to do that—and to build on the knowledge that generalists are already offering. It’s a wonderful symbiosis that brings expert knowledge into the conversation and makes it accessible.

Jan Thomas: Yes. And maybe just to recap before we talk about what data is available here and can be pulled up via MCP: You just mentioned that you’ve been involved in around 100 transactions. What lessons were particularly relevant to you and led you to conclude that another layer of information is needed—to help founders avoid getting into situations like the ones you’ve perhaps seen a few times before?

Björn Rieckhoff: I think there are numerous points that could be mentioned here. I’ve approached the whole topic from the perspective of a founder who is facing this fundraising challenge for the first time. So we’re not necessarily talking about the serial entrepreneur who’s raising funds for the umpteenth time—side note: They can benefit from this, too, of course. But someone dealing with this for the first time needs a starting point somewhere: What investors are out there, how do I categorize them, and so on. When I think back to my time at Cavalry—I started working there in 2016, and AI wasn’t a big topic yet. Back then, there were still lists maintained entirely by hand that were tracked somehow in the CRM. And then, as a VC, you’d go in to support your founders and set filters by geography, stage, industry, and the like to narrow things down to a shortlist.

Jan Thomas: So, basically keyword-based.

Björn Rieckhoff: That was actually based on keywords, and you still didn’t have a qualified list in the sense of: a deep dive into their portfolio—what’s their actual allocation behavior? Instead, you were relying on hearsay and whatever information you could find here and there. Of course, the whole thing has evolved; by now, people have gotten a bit smarter about how to do it. Side note: I still see VCs who operate exactly the same way. And, surprisingly, even in advisory engagements—when a founder asks within their investor network, “Who should I approach for the next funding round?” unfortunately, they still get an email with five names—essentially unqualified ones that are just top of mind—without anyone having done any research or vetting. And then the founder sits there and starts doing their own research using the standard tools. Of course, you can also turn to Crunchbase, Carta, and the like, or even use a fundraising CRM. All of that is perfectly fine. I just always get the impression that, on the one hand, it’s a pricing issue because these tools are often priced for later-stage funding. And on the other hand, it’s also a question of—as you said earlier—where does the knowledge actually reside, and where is it manageable? I simply think: I don’t need a separate tool to manage my fundraising; rather, I want to keep it dynamically accessible during conversations. And then there isn’t a lot of back-and-forth—instead, it gives you a qualified shortlist of investors that you can expand as needed, conduct in-depth research on, and so on.

---

Jan Thomas: What does “qualified” mean to you? What does that actually entail? You just said that typical lists don’t provide a deep dive into the portfolios. Is that what qualifies them? Or is it more about feedback from the market? What’s important?

Björn Rieckhoff: A whole lot of different things—and it’s a dynamic list that’s constantly expanding. I started out in a very traditional way: systematically organizing a founder’s business plan and tagging it in the same way an investor would. Basically, you have an intake form—which isn’t on the website either, but takes place during a sort of intake conversation with the client, where the founder explains: What’s the problem, what’s the solution, where do I actually stand, what am I looking for in an investor, what expertise do I need, who’s my target audience—is it enterprise, is it B2C, or something similar. From there, I then cross-reference this with the investor database I’ve built. And that’s where a deterministic matching process takes place first. This is repeatable—and that’s better than getting different answers from two different clients; I get the same results every time. Then it goes deeper, focusing more on current investment behavior: Which companies have received funding in the past? Can trends be identified? Can you tell whether there’s a lead investor or not? At the team level: Who, if anyone, would be responsible for the relevant area in order to identify a specific point of contact? And here I’ll also say: I’m less concerned with building a sales tool. I don’t want to do that—there are already plenty of those. My goal is to provide the information necessary for outreach.

Jan Thomas: What distinguishes a good investor from a bad one? It’s not so much a question of “compatibility” as it is: What are the things you’d look out for when you’re sitting down with an investor? I’d also like to ask about this: Many people say that the best way to connect with an investor is through personal contact, either directly or through mutual acquaintances. And that has a lot to do with not only understanding the VC’s focus but also knowing what kind of people they are, as individuals. And it may well be that the person you end up dealing with isn’t necessarily the one you click with—to put it mildly.

Björn Rieckhoff: That’s absolutely true. And I have to say: That’s a limitation I’ve made very deliberately. It’s true that what’s in there is—if you will—my nuanced understanding of and knowledge about the industry. My personal opinion also factors into the assessment of how it all comes together—how this shortlist is generated—into the evaluation algorithm, so to speak. However, that’s not a personal opinion when it comes to whether an investor is any good or not. It’s simply neutral. There’s no judgment involved.

Jan Thomas: Because that does happen sometimes. I’m familiar with projects like that—I don’t know if there’s a platform that’s actually pulled it off—that try to incorporate founder feedback. It’s a bit like Glassdoor or Kununu for employers, but essentially for VCs. Or even for business angels. Maybe we can also talk about what kinds of investors you find on your platform and which ones you don’t—because business angels are perhaps even less transparent than the VC market.

Björn Rieckhoff: That has actually been a focus area for me lately as well—expanding our coverage and adding depth. What’s important to me is conveying factual knowledge. It shouldn’t just be the grapevine among founders that’s reflected here. If someone doesn’t reply to an email for two weeks, then they simply aren’t replying to an email for two weeks. That might become apparent later in their investment behavior or in the round’s allocations, but it isn’t necessarily explicitly mentioned on Nephilia. My goal is rather to say relatively early on: “Don’t approach that person.” But that’s based on factual information—in the sense that “this investor simply doesn’t invest in your sector”—rather than on the assumption that “the investor might not be responsive.” The point is to provide the necessary information so that the founder can make their own decision.

Jan Thomas: When you look at employer review sites, it’s a bit like Google Maps or restaurant reviews: the people who speak up are either totally thrilled or really annoyed. The vast majority in the middle—who could provide the most objective picture—don’t weigh in. So I understand that. Your goal is probably also to help founders make the right decision while saving them time. Because if I end up making a relatively large number of calls to the “wrong” ones—in quotes, because maybe not everyone is a good fit for me—and then have to follow up with phone calls and chase them down, that’s obviously something to avoid.

Björn Rieckhoff: That’s basically one of my value propositions: that I’ll save you time on the operational side of fundraising. Whether that leads to greater success or not, I can’t say. There are so many factors at play that you have no control over—it wouldn’t be fair to claim otherwise. But here’s what’s true: You proactively avoid those conversations where someone—after you’ve maybe had two or three good points of contact—says: “Oh no, we’re not doing that at all right now.” And then you get a completely generic response. I’m not saying that won’t happen to you with Nephilia—it’ll still happen from time to time. But I assume you’ll be able to filter out a lot of that by at least having the relevant partners in place from the start. Side note: To be honest, I find this really exciting from a VC perspective, too, because there’s mutual interest in ensuring that the underlying information I’m working with is accurate —because, in a pinch, that gives the VC a better funnel instead of having to process the bulk of investment opportunities that don’t fit the focus anyway.

Jan Thomas: That would be ideal. At the same time, I know a great many VCs who, here on the podcast or at events, are always boasting about how many deals they see in a year. That’s the kind of metric where you think: Okay, you look at 10,000 deals but only make ten investments—congratulations on your efficiency. That’s why it would be ideal for both sides if you knew from the start who you need to approach, and the VC only received relevant deal flow. I’m not exactly sure if that’s the KPI the VC really wants to optimize.

Björn Rieckhoff: Nope. Personally, I’ve always been against prominently listing a KPI like that because I found it completely irrelevant—but there were also differing opinions within the fund. Fair enough. But as I said: I built Nephilia from a founder’s perspective to help founders through the process. It’s not a tool that provides deal flow for VCs—not even in the long run. In a way, I saw it as a conflict to monetize on both sides. Ask me again in three years—maybe I’ll think differently by then—but it’s an interesting position.

Jan Thomas: No, I think that’s important. I’d start by agreeing with you on that: you shouldn’t be a lawyer for both sides, but rather say, “I’m focusing on one side and trying to optimize things for them.”

Björn Rieckhoff: Exactly. And I think that as someone who has seen and understands these dynamics, you can pick up on a lot of signals that you wouldn’t be able to capture by asking Claude or ChatGPT a general question. For one thing, the depth is different. Second, there are a great many dynamics that I myself only discovered while building the whole thing—that is, while defining the MCP’s behavior—where you can delve much deeper. Not just the deployment behavior and coverage of certain geographies, industries, target groups, focus areas, and the like, but also: What kind of LP base do you have, if anything? That isn’t implemented yet because it would still be a bit too superficial to run a valuation on it. But looking ahead, I can very well imagine that this “Direct Access”—which manifests itself in this way—could play a role in qualifying a fund for the founder. And it’s not at all about saying one ranks higher than another, but rather about providing the founder with an objective note: “Look, you’re in a sector where LP XY could be a great contact for you—who, by the way, is invested in such-and-such fund”—simply to highlight opportunities for access.

---

Jan Thomas: Do you have an ICP right now—the ideal founder or founding team to start with, or the one you feel most comfortable with at the moment?

Björn Rieckhoff: I think it’s a Seed-stage company.

Jan Thomas: From Germany?

Björn Rieckhoff: I’d say Central Europe. The corpus—the database—actually covers prominent investors in Europe and the U.S., but personally, I’ve currently placed a very strong emphasis on Europe in terms of the depth of the information. That’s also what drives me: I want to have in-depth coverage in Europe. The DACH region certainly has the most in-depth coverage. This has to do with how the data sources are connected: Coverage of commercial registers in Southern Europe isn’t currently in-depth enough for me to program it. But the commercial register, for example, is already included, and then you build on the hard facts from the commercial register, which is of course more reliable than media reports. ICP, I’d say, is also about that—I’ve noticed often enough: For pre-seed founders, it’s about willingness to pay on one hand, and on the other, it’s about gaining their own experience with fundraising in the first place. Someone who’s never done it before probably doesn’t realize just how painful it is. And maybe they have to go through that cycle of “I’ve spoken to a lot of investors, unfortunately none of whom were relevant” to realize that.

Jan Thomas: Sorry to interrupt—I don’t think I’d rush out and talk to my top five dream investors right away anyway. Instead, I’d probably take a few knocks first from, let’s say, ten “throwaway” investors on my list, just to practice my pitch.

Björn Rieckhoff: Of course you can do that—it’s entirely up to you. You can also flip the list I’m giving you—or just ask Nephilia, “Please find me some that don’t fit at all.”

Jan Thomas: Nah, but I don’t think it would be smart to just jump right in and totally mess up on your dream date or something like that. I do agree with you there.

Björn Rieckhoff: Of course, there’s an entire aspect of this that you can’t handle yourself. That’s why I decided to start by building a highly specialized information platform for this area rather than providing end-to-end coverage, as you sometimes see— “this is your agency-style fundraising framework; we’ll build your pitch deck, identify the investors, send out the requests, and then handle the operational process.” I explicitly didn’t want to build that because I believe that when it comes to such a central question as “Who should I actually take money from?”, the founder will always have a very, very strong opinion—and the willingness to hand that over completely will be limited. My experience is this: Despite all the tools out there, I haven’t yet met a founder who didn’t write the outreach emails themselves—or at least served as the “human in the loop” to redraft them. This simply shows that the stakes are too high to automate this process 100 percent. But of course, you should work from a reliable data set—and that’s what I’d like to provide.

Jan Thomas: And this data set—let’s go over it again to look for any potential blind spots: I gather there’s a certain depth in Europe and Central Europe, and a certain breadth in the U.S. In other words, it’s sector-agnostic, as I understand it. So a startup in the Seed stage wouldn’t necessarily have to come from a specific sector?

Björn Rieckhoff: Nope, not that. Of course—just look at the general investor landscape out there—there are more general-tech funds than there are biotech and pharma funds.

Jan Thomas: Yeah, but that’s also important—that people can find them through you.

Björn Rieckhoff: Exactly. What was important to me for the launch—I even called it “high-profile” myself—were funds and investors that historically have a good reputation and, as a result, enjoy good press coverage. They were the main focus. I didn’t want to risk the embarrassment of someone on the Midas List not appearing in Nephilia. That was basically what it came down to for me.

Jan Thomas: Yeah, absolutely. That’s exactly the example I’m getting at, because all of this is in hindsight—these are all historical analyses. Let’s take AI, for example: AI came onto the scene at some point, and almost no VC really had it on their radar. I mean, Fly Ventures here in Berlin has been out there for years saying this is coming, but the ChatGPT moment was the turning point for the whole scene. It’s the same with defense tech—that’s been changing gradually, first dual-use and then full-fledged defense tech. What I’m getting at is: How do you handle it when new topics emerge that, for all intents and purposes, haven’t been invested in yet? How do you deal with that?

Björn Rieckhoff: Basically, you can only use the data that’s currently available as a basis. Of course, I keep an ear to the ground myself for the latest developments and incorporate that into the database, but as an individual, I don’t have the coverage I’d need to make it representative. Of course, you have to recognize that the things being tracked—currently, robotics and foundational AI are also tagged—evolve over time. I expect this to have a certain dynamic once the tool is in use, because you’ll also have a kind of feedback loop based on the founders who use it: Which funds are investing where, and what are the data points?

Jan Thomas: So you do that, too—a feedback loop where users provide data?

Björn Rieckhoff: Yes, MCP works pretty well for collecting this kind of feedback. The question is how willing each founder is to share that information with the tool. I wouldn’t say that Nephilia is actively tracking that—not at the moment. It might be possible to do that at some point in the future, based on how the server behaves. But no one is forced to reveal how their fundraising is going. Of course, looking ahead, this does give you insight into current trends. But of course, you’re running a bit behind, based on current deployment behavior, which is used as the basis for evaluating the funds.

---

Jan Thomas: Yeah, because there are topics that are relatively straightforward—e-commerce or food and beverage or something like that. In those cases, you’ll figure out pretty quickly who’s investing in them and who isn’t. But as soon as you get into more specialized areas—we both talked about Oura a while back. If someone comes along now and says, “I’m launching an Oura competitor out of Germany,” there are probably very few VCs in Germany who you’d immediately know would take a look at it, because there hasn’t been any investment in that area here at all. Wearables and things like that are a bit of a blind spot. And if you take that further—I don’t know how you view it, but this topic of singularity and human-computer interfaces is going to become increasingly prominent. It’s probably still a taboo topic for most VCs right now, but if you look at Neuralink or something like that: sooner or later, these topics will come up. And then, using common sense—it’s not so much about matching buzzwords—I wouldn’t even know how to proceed so as not to miss the mark a hundred times. Should you talk to a generalist like HV Capital first, or to pure deep-tech VCs?

Björn Rieckhoff: Yes, that’s a good example. As you say, this isn’t something a fund currently defines as part of its mandate. It’s constantly evolving. I think it helps to add that extra layer of information—the “Why We Invested In” posts, all those blog posts, essentially the industry statements.

Jan Thomas: And you can tell that from that, too?

Björn Rieckhoff: Of course, it’s not as if I’ve turned everything floating around in my head into a fully developed product by the time of the launch. Not at all. But I would think that’s definitely a layer that could be added in the future. I also deliberately didn’t position it as purely a fundraising tool; rather, the long-term vision is for it to be an information service for founders. That might even mean saying, “Look, your startup looks promising—but in all fairness, the current trend and the investment behavior of general-tech funds are, unfortunately, working against you right now.” It’s not at all about telling the founder that their idea is no good, but rather about setting realistic expectations to support the conversations the founder is having. I’ve actively noticed this in my consulting work and, even before that, in portfolio management: You can only influence the decisions of other funds to a limited extent—side note: you rarely know in advance, either. But you can actively tell the founder: “In this situation, you should talk to 50 funds rather than just five, simply because the idea might not really resonate.” That’s exactly how it should play out: the founder should find guidance and not be left standing there like a deer in the headlights, unsure of how to approach the issue.

Jan Thomas: Actually, having a scoring system like that at the beginning would be pretty cool. It would tell you, “Hey, what you’ve set out to do with this fundraising is going to be quite a challenge.” From 0 to 100—that would be a surefire home run within three days, versus having to plan for three months and 50 VCs.

Björn Rieckhoff: But from my experience, I know that’s where people tend to make mistakes. Even with things where you’d say you’ve exceeded all the KPIs you’d need for a Series A—I’ve had experiences where—I don’t want to say I’ve seen the horses throw up—but it still takes a surprisingly long time, because you always find something if you’re not 100 percent convinced.

Jan Thomas: Nah, I agree: If I ask for more data now—like how things work in the more complicated cases—and you say you can also analyze blog posts and things like that—if I were you, I wouldn’t even do that at the beginning. You have to take a Pareto-style approach: I can cover the majority of cases with this level of effort, and if it gets into specialized areas, then probably only once the tool is established, once that feedback loop is in place.

Björn Rieckhoff: Exactly. And, of course—to put it simply—it shouldn’t reward charlatans. In other words, someone who just talks a lot.

Jan Thomas: On the VC side? Is there such a thing?

Björn Rieckhoff: Anyone who makes a lot of information indexable naturally benefits from being ranked higher somewhere. But that’s classic SEO and GEO behavior, and in a way, it’s also reflected in tools like Nephilia—that’s just how it is. Someone who, like Hedosophia, doesn’t say a single word to the outside world—well, in that case, there’s less to go on, except for their active deployment behavior. You can still count them as a high-profile investor.

Jan Thomas: Interesting. But that’s probably one of the few examples that’s really kept under wraps, where very little is known.

Björn Rieckhoff: Yeah, I’d say so too, especially right now.

Jan Thomas: So, let’s talk a little more about the risks—we’ve talked a lot about the strengths so far. In the age of AI, the topic of data—especially a data-driven business—tends to have a rather uncertain future. Aren’t you worried that what works today will be common knowledge in two years—that almost everyone will be able to access this information through Claude and the like?

Björn Rieckhoff: This is already the case now—not just in two or three years: If you were to analyze the data available from Crunchbase, Carta, or, if necessary, a commercial registry API provider in a structured way, I think you’d get pretty far. But I have to say myself: While building the entire database and this data foundation, I realized that here and there, I always have to make pivotal decisions—in the sense of: Do I go left, or do I go right?—and that’s where my background really helps me. I built it primarily for myself and then expanded its scope. And sure, the data foundation is an issue. If you were to claim you have the broadest one, you’d just be playing catch-up—you’d be spending all your time building integrations and still be lagging behind. What’s interesting is the use case: that you get the insights sooner than others when in doubt. That you realize: “Hey, Cherry is investing here and here right now, but maybe not there.” Establishing this feedback loop as a flywheel can be exciting. Plus: Personally, I’m not interested in becoming an aggregator, because I always lose out there. It’s more about bringing in that level of judgment: What do I do with the information? And how do I make the best possible use of it for what is—arguably—a very niche use case, when viewed across the broader economy: pre-seed through Series B funding rounds for tech companies. Incredibly niche.

---

Jan Thomas: Interesting. Another point that probably comes into play: the topic of seed-strapping. Do you see that as something that’s changing things for you? We’re talking about VCs right now—but maybe the whole VC thing isn’t even relevant for a lot of people anymore; instead, they’ll say: I’ll put together an early-stage funding round of one or two million, and that’s it. And with that, I’ll build an AI wrapper or some kind of tool—just like you did; we can talk about exactly how to do that in a moment. I think it’s totally crazy how we’re all suddenly becoming coders and product designers. You built this all on your own. A few years ago, that wasn’t even possible—it was unthinkable. Back then, you would’ve needed seed funding. And many people who are launching an app now say: “I’m building this, but I still need a little marketing budget or two or three people on the team”—and that’s it; we’re actually supposed to be able to turn a profit with just that. Does that change how you approach your tool or your perspective on it?

Björn Rieckhoff: It’s certainly a matter of perspective. Of course, it’s a tool geared toward the venture capital market. If the venture capital market in general is struggling and everyone is building their business without VC funding, then I’ll be left out in the cold, of course.

Jan Thomas: Seed-strapping doesn’t have to be completely at odds with VC—there are probably both paths. But as a founder, it’s naturally the case that once you go down that path, you’ll eventually have to make an exit—following the logic of VC.

Björn Rieckhoff: One of the most common conversations I have very, very early on in my advisory engagements is the fundamental question: From whom should I raise money—and should I even do so at all? Even though I’ve spent my entire professional life in the VC sector, I’m not necessarily an advocate for the idea that this is the right fit for every business. Not at all. If you’re able to build a business without external funding—all the better. Great. But having that option—at a certain stage, due to competitive pressure, growth ambitions, and the like—to pour more money into a functioning business model—I’ve always personally found that to be quite smart. There are some very, very good examples of this. And I also believe that when it comes to exits: Founders who have been very prudent with investor funds aren’t necessarily worse off financially than those who’ve taken in endless amounts of money.

Jan Thomas: We have a few prominent examples—something like Bending Spoons probably wouldn’t even exist otherwise. And maybe you could briefly share your insights again: If someone comes along with an idea—say, an MCP project or some kind of data-related concept—and says, “I’d like to build something like that,” how long does it take to set something like that up? We talked about this a few months ago, but I don’t know your timeline, man-hours, and things like that—or tokens; that’s probably the right metric.

Björn Rieckhoff: That’s definitely doable. For me, the real question isn’t whether it can be implemented—from a coding standpoint, it’s certainly possible—but rather: What am I doing here, and why am I doing it exactly the way I do? That’s the knowledge that’s important for figuring this out, because no AI can take that away from you. Of course, you can listen to every recommendation the AI makes. But then, I think, you’ll end up taking three wrong turns or ending up with a setup that you don’t really want—at least not right now. For me, the much larger part of “How long does this take?” was actually grappling with the subject matter. And then the question is: Where do you start counting? For me, if you will, it’s based on ten years of work experience in this field. As for the actual implementation—I worked on it alone—I’d estimate it would take about a month to a month and a half for the MCP itself. But that’s mostly authentication, security, and server behavior. The much bigger task is the depth of the data and the engine in between that asks: What do we do with all this information?

Jan Thomas: Wow. I think it looks great—I love the idea. Finally: Feel free to do another quick promotional segment. It’s already available for use—people can sign up. Maybe also say a few words about the pricing—how you envision it—and perhaps mention who shouldn’t use it, but especially who should.

Björn Rieckhoff: Who should use it—exactly. Nephilia, available at nephilia.ai, is an information service that helps you, as a founder, raise funds from the funds that are actually relevant to your field. You integrate it into your own assistant via an MCP or a connector—Claude, OpenAI—and it’ll likely be listed as an official plugin soon; I’m working on that right now. I’ve decided to offer a flat rate per funding round. This isn’t a subscription, but a fixed price per funding round: starting at 900 euros for Pre-Seed, up to 2,500 euros for Series A. It’s somewhat tailored to affordability—and, to be honest, also with the idea of supporting Pre-Seed companies. And as of the recording of this podcast, it’s in public beta, so it’s freely available. You can buy it.

Jan Thomas: Cool. Like I said, I think it’s a cool idea. But when you say “Pre-Seed”—since these terms are always a bit vague, with some calling it Series A and others a Seed round—is that determined later based on the planned fundraising amount?

Björn Rieckhoff: Yes. I generally define Pre-Seed as up to one and a half million in funding volume, Seed funding round as one and a half to seven—that’s how I’ve classified it for myself—and Series A as seven to twenty. Those are the ranges. If you stay within these ranges, the pricing is the same. If I start with one million and then have Nephilia help me find lead investors for a 5-million funding round, the system actually responds appropriately and asks for a top-up to the next higher category. This isn’t meant for AI startups doing a 50-million seed round—they’re not what we mean by “seed.”

Jan Thomas: What do you think—in three months, where will your strongest market be? Germany? I sure hope so.

Björn Rieckhoff: I’m personally quite fond of Central Europe as well. I think the investor side will be exciting: I actually believe that a lot of U.S. funds that also invest in Europe will be approached quite frequently. After all, we’re seeing more and more companies raising funds from U.S. funds early on—simply because an understanding of AI, the level of ambition, and an appreciation for the corresponding growth rates of these businesses are perhaps a bit more deeply ingrained over there than they are here.

Jan Thomas: Yeah, I was asking more with that in mind: Sure, that’s what we’d all like to see, but where is the problem the biggest? You could have said: In Portugal, Spain, or Poland, there are an incredible number of companies that actually want to connect with Central European funds but have less transparency about the market than the Germans do. That’s why I thought that.

Björn Rieckhoff: Yeah, that’s interesting. I can’t really tell you right now, but I hope that in the future I’ll have some data to answer exactly those kinds of questions. When it comes to market research, it’s a really exciting position to be in.

Jan Thomas: Cool, Björn, I’m keeping my fingers crossed for you. That’s really cool. I love anything entrepreneurial, and I think this is the start of a new project. Did we forget anything important?

Björn Rieckhoff: Not on my end, no.

Jan Thomas: Cool, so as I said: the link is in the show notes. And we’d be thrilled if this goes viral—then we’ll do an update at some point, or we can briefly talk about it every time we chat.

Björn Rieckhoff: I’d be happy to do that, yes. Thanks, Jan.

Jan Thomas: See you later.

---

About Björn Rieckhoff

Björn Rieckhoff is the founder of Nephilia and a business angel with ten years of experience in early-stage venture capital. He helped build Cavalry Ventures as its first employee and later became a partner of the fund. Nephilia grew out of around 100 transactions and board seats from seed to Series B: an information service that delivers investor selection for a founder's round directly inside their own AI assistant. Alongside it, he remains a sparring partner for entrepreneurs.

About Startup Insider

Startup Insider is the industry portal for the startup scene in the DACH region. It covers news from all regions and industries, along with an overview of key players and events in the German-speaking startup world.

Links

• Nephilia – Informationsdienst für Gründer (Public Beta)
• Notiz zum Start von Nephilia auf bjolo.de
• Model Context Protocol (MCP)
• Oura erreicht 11-Milliarden-Bewertung: Wie weit trägt ein einzelnes Produkt? (im Gespräch erwähnte Episode)