SignalFire Automates Founder Support to Secure 100% Follow-On Wins
Ai4
Replacing manual venture services with automated data platforms prevents portfolio neglect and secures pro-rata access to competitive follow-on rounds.
## Background
Published on YouTube by the Ai4 channel on August 28, 2025, this presentation features Ilya Kirnos, Co-Founder and CTO of SignalFire. According to the page description, the talk examines why capital alone is insufficient for venture differentiation, detailing how SignalFire engineered scalable software value-adds and operational infrastructure to effectively serve founders and scale portfolio support.
## TL;DR
Ilya Kirnos breaks down how SignalFire was architected from first principles as a technology company to overcome the manual, artisanal limitations of traditional venture capital. The firm operates an enterprise data platform tracking 80 million companies and 600 million profiles to power early sourcing, rigorous competitive diligence, and automated post-investment support. By delivering proprietary recruiting and go-to-market software directly to portfolio companies, SignalFire eliminates portfolio orphanage and maintains a 100% pro rata win rate in competitive follow-on rounds. Case studies of Horizon3.ai and EvenUp demonstrate how combining scalable programmatic tools with human-in-the-loop executive advisory creates a compounding flywheel for startup growth and investor returns.
## Atmosphere & style
The presentation carries a crisp, pragmatic, engineering-first tone that systematically deconstructs legacy venture practices with analytical clarity. Kirnos presents with the calm, measured authority of a veteran systems architect, blending personal anecdotes from Google and Silicon Valley with concrete operational frameworks. Listening to the talk feels less like a standard promotional pitch and more like an insider technical post-mortem on modernizing an outdated financial industry.
## Key takeaways
- [01:15](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=75s) Traditional venture capital historically operates with high manual overhead, intuition, and personal networks, lacking the proprietary technology and automation standard in other software-driven sectors.
- [00:04](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=4s) SignalFire was designed from inception as an engineering-driven firm, currently managing approximately $3 billion in AUM with an emphasis on Seed through Series B investing.
- [02:44](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=164s) Venture success relies on a four-pillar lifecycle—Sourcing, Diligence and Selection, Winning the Deal, and Post-Investment Support—anchored by a reputational flywheel driven by founder advocacy.
- [05:25](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=325s) SignalFire built an enterprise data engine tracking roughly 80 million companies and 600 million professional profiles to detect real-time signals like engineering hiring spikes, app download velocity, and web traffic inflection points.
- [05:25](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=325s) Despite deep algorithmic sourcing and benchmarking, early-stage VC remains an inherently human-in-the-loop discipline requiring partners to evaluate founder vision, resilience, and recruiting charisma.
- [07:34](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=454s) The proprietary data infrastructure used for deal sourcing is exposed directly to portfolio founders at no cost to power go-to-market prospecting, competitive intelligence, and technical recruiting.
- [07:34](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=454s) Operational platform support dynamically shifts with market cycles: the 2021 bull market demanded heavy technical hiring assistance, while higher interest rates shifted founder focus to capital-efficient GTM acceleration.
- [09:35](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=575s) Talent needs naturally split into self-serve programmatic tools for volume individual contributor hiring and high-touch in-house advisory for executive leadership placement.
- [12:50](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=770s) Traditional manual platform teams suffer from linear staff bottlenecks, forcing firms to triage and orphan struggling startups during non-linear pivot phases.
- [12:50](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=770s) Automating baseline platform services through software ensures universal portfolio access, eliminating portfolio orphanage and enabling a 100% win rate on competitive pro-rata follow-on rounds.
- [16:08](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=968s) For cybersecurity portfolio company Horizon3.ai, SignalFire leveraged its executive network to recruit roughly 75% of the C-suite and evaluate engineering leadership candidates.
- [19:14](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=1154s) For legal intelligence startup EvenUp, which began with non-technical founders, SignalFire acted as the primary recruiting engine through Series B, helping scale the company toward a valuation well over $1.25 billion.
- [20:52](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=1252s) Building scalable software infrastructure within VC shifts the firm from a passive capital allocator to an operational partner, compounding advantages for both founders and limited partners.
## Detailed walkthrough
### [00:04](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=4s) Ilya Kirnos and the Origins of SignalFire
Ilya Kirnos brings an engineering-first perspective to the venture capital ecosystem. After completing an undergraduate double major in Computer Science and Mathematics on the East Coast twenty-five years ago, he moved directly to Silicon Valley to begin his career on Oracle's database team. In 2003, Kirnos transitioned to Google, where he spent nine years architecting core infrastructure across AdWords, Gmail, and Gmail monetization. In mid-2012, he departed Google for a startup that was quickly acquired by Twitter; he left prior to the acquisition to co-found SignalFire as its technical co-founder and Chief Technology Officer.
```mermaid
flowchart LR
accTitle: Ilya Kirnos Career Path to SignalFire
subgraph Career["Engineering Background"]
A["Oracle\nDatabase Team"] --> B["Google (2003–2012)\nAdWords & Gmail"]
B --> C["Startup\nAcquired by Twitter"]
end
subgraph Venture["Venture Capital"]
C --> D["SignalFire\nCo-Founder & CTO"]
end
```
SignalFire was established as the first venture capital firm designed and engineered from the ground up like a modern technology company. The firm currently manages approximately $3 billion in assets under management (AUM), with its current active fund totaling roughly $1 billion. While SignalFire participates across multiple stages up to Series C, its primary investment engine focuses squarely on Seed, Series A, and Series B rounds.
- **Founder Background:** CS/Math undergrad; Oracle database engineer; 9-year tenure at Google (AdWords infrastructure, Gmail monetization).
- **Firm Scale:** ~$3 billion total AUM; ~$1 billion current vehicle.
- **Stage Focus:** Primarily Seed through Series A and B, scaling up to Series C.
---
[01:15](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=75s) The Irony of Venture Capital and First-Principles Thinking
A glaring paradox sits at the heart of traditional venture capital: while venture capitalists allocate billions into cutting-edge technology companies, the VC sector itself has historically operated in a deeply tech-deficient, manual manner. When SignalFire was founded twelve years ago, nearly every established fund operated purely on manual services, intuition, and personal networks. Kirnos often framed this as a thought experiment: if a VC firm were forced to pitch other venture capitalists to raise capital, it would fail the very diligence tests it applies to startups. Firms had no proprietary software, zero automation, and virtually no product differentiation beyond legacy brand reputation.
When SignalFire proposed applying data engineering and software infrastructure to venture capital—mirroring transformations across virtually every other industry vertical—the idea was met with widespread industry skepticism. The prevailing dogma held that venture capital was an artisanal, "apprentice industry" governed by art rather than science. Skeptics questioned how data could possibly evaluate early-stage seed companies that possessed nothing more than a founding team and an unproven idea. SignalFire addressed this by breaking venture investing down to its foundational first principles.
```mermaid
flowchart TD
accTitle: Four Pillars of VC and the Reputational Flywheel
subgraph Pillars["Core Operating Pillars"]
S["1. Sourcing\nIdentify companies early"] --> D["2. Diligence & Selection\nBenchmark 1-of-1 vs. 1-of-many"]
D --> W["Intermediate: Winning the Deal\nSharp calling card & founder value-add"]
W --> P["3. Post-Investment Support\nRecruiting, GTM, & Competitive Intel"]
end
subgraph Flywheel["Reputational Flywheel"]
P --> F1["Portfolio Founder Reference Checks"]
F1 --> F2["High Founder Advocacy"]
F2 --> W
end
```
---
[02:44](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=164s) The Four Pillars of VC and the Reputational Flywheel
SignalFire deconstructed the venture lifecycle into four core pillars required to consistently outperform the market:
1. **Sourcing:** Identifying nascent startups before they appear on the broader market's radar and establishing relationships as the first institutional investor.
2. **Diligence and Selection:** Evaluating whether a prospective startup is truly "one of one" or merely "one of many" in a crowded vertical. Kirnos illustrates this with the historical ride-sharing boom: an investor looking at Uber, Lyft, and eight other competing startups needed rigorous mechanisms to pick the enduring winners while avoiding the forgotten competitors.
3. **Winning the Deal:** High-potential founders inevitably receive multiple competing term sheets; a fund must offer a sharp, differentiated "calling card" to become the founder's investor of choice.
4. **Post-Investment Support:** Accelerating portfolio growth through operational scale, which directly fuels the firm's overarching reputational flywheel.
When founders evaluate prospective lead investors, their primary due diligence involves backchanneling with current portfolio founders. If a fund reliably delivers on its operational promises, those founders become enthusiastic advocates. This advocacy directly increases the firm's win rate on subsequent competitive deals, spinning the flywheel faster and compounding the firm's brand equity over time.
---
[05:25](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=325s) Big Data, Machine Learning, and Sourcing Architecture
To power early sourcing and competitive benchmarking, SignalFire constructed an enterprise-scale data platform that tracks approximately 80 million companies in real time alongside over 600 million individual professional profiles. The engineering platform ingests fragmented data across disparate external silos, cleanses and normalizes the incoming streams, and resolves classic entity-resolution challenges at scale. Proprietary machine learning models continuously evaluate these datasets to detect breakout momentum signals:
- Accelerated high-caliber talent migration and key engineering hires.
- Rapid revenue trajectory inflection points.
- Spikes in web traffic and digital engagement.
- Surges in mobile application store download velocity.
```mermaid
flowchart TD
accTitle: SignalFire Data Pipeline and Human-in-the-Loop Funnel
subgraph DataLayer["Data Platform (80M Companies / 600M Profiles)"]
D1["Disparate Silos & Web Data"] --> D2["Data Cleansing & Entity Resolution"]
D2 --> ML["ML / AI Models\nTalent Flow, Revenue, Web Traffic, App Downloads"]
end
subgraph DecisionLayer["Human-in-the-Loop Selection"]
ML --> Sigs["Momentum Signals & Benchmarking"]
Sigs --> Human["Investment Team Diligence\nVision, Passion, Selling & Recruiting Ability"]
Human --> TermSheet["Investment Decision"]
end
```
Despite extensive automation, Kirnos emphasizes that venture capital remains fundamentally a "human-in-the-loop" problem. Algorithmic signals bubble up the top percentile of opportunities, but human investment partners must lead qualitative diligence. Data cannot replace direct human evaluation of a founder's vision, charisma, resilience, salesmanship, and ability to attract elite talent. Once a target is engaged, the platform provides immediate comparative benchmarking against its closest peer set across core operating dimensions.
---
[07:34](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=454s) Repurposing Platform Data for Portfolio Operations and Market Cycles
The identical data infrastructure that drives investment sourcing is exposed directly to portfolio founders free of charge as a decisive value-add to win deals and accelerate portfolio growth. Founders leverage the platform to gain granular competitive intelligence on market rivals, conduct deep go-to-market (GTM) prospecting, and solve mission-critical recruiting challenges.
```mermaid
flowchart LR
accTitle: Macro Market Shifts and Portfolio Needs
subgraph Peak2021["2021 Bull Market"]
A1["Capital Abundant\nLoose Spending"] --> B1["Primary Bottleneck:\nTechnical Recruiting"]
end
subgraph Modern["Higher Rate Environment"]
A2["Tech Layoffs\nCapital Discipline"] --> B2["Primary Bottleneck:\nEfficient GTM & Prospecting"]
end
```
Kirnos highlights how portfolio needs dynamically shift across macro market cycles:
- **The 2021 Market Peak:** Capital was abundant and customer spending was loose. Go-to-market was rarely a constraint; the primary bottleneck to company growth was hiring top-tier technical and engineering talent in an intensely competitive labor market.
- **The Higher Interest Rate Regime:** As tech giants executed layoffs and capital contracted, talent availability eased while the mandate shifted from "growth at all costs" to capital efficiency. SignalFire adjusted its platform support toward deep GTM acceleration—enabling companies (such as B2B data platforms) to query the database, map exact corporate tech stacks, and systematically generate high-conversion prospect lists.
[09:35](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=575s) Applying Data and AI to Post-Investment Support: Talent, GTM, and Intelligence
SignalFire applies proprietary data and software to venture capital with a heavy emphasis on post-investment support. All early-stage portfolio companies face massive operational hurdles across three core pillars: talent acquisition, go-to-market (GTM) acceleration, and competitive intelligence.
Recruiting needs naturally bifurcate into two operational models:
- **Volume Recruiting (Individual Contributors):** As seed-stage startups grow to Series B and C, hiring shifts from adding one or two engineers to needing a scalable "machine for getting talent in the door." SignalFire services this via scalable, self-serve software products.
- **Executive Recruiting (Leadership):** Hiring point-in-time executive roles (such as a Chief Revenue Officer or Chief Marketing Officer) requires high-touch advisory. SignalFire maintains an in-house executive search team staffed with talent alumni from Facebook, Google, and Waymo.
For go-to-market acceleration, SignalFire provides specialized prospecting datasets designed specifically for early-stage technical startups, functioning like a "ZoomInfo on steroids" without attempting to compete directly with broad-market commercial vendors. For competitive intelligence, data pipelines deliver granular market analysis to help founders allocate capital efficiently.
```mermaid
flowchart TD
accTitle: SignalFire Post-Investment Support Architecture
subgraph Talent["Talent Acquisition"]
A["Volume IC Hiring"] --> B["Self-Serve Software Platform"]
C["Executive Search"] --> D["White-Glove In-House Team"]
end
- **Uber vs. Lyft Case Study:** During Uber's market battles with Lyft, market dynamics varied drastically by geography (e.g., Lyft maintained high strength in Los Angeles compared to New York). SignalFire used granular data to slice and dice rider revenue, retention rates, and customer lifetime value (LTV) across specific metropolitan areas, allowing Uber to optimize marketing and customer acquisition spend.
---
[12:50](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=770s) The Pitfall of Traditional VC: Eliminating Portfolio Orphanage
In traditional venture capital firms, platform support is delivered entirely manually. As a consequence, portfolio expansion directly degrades service quality: a team of 10 recruiters can provide deep support when managing 10 companies, but the model collapses when the portfolio scales to 100 companies. Traditional firms are forced to triage resources, allocating human capital exclusively to top-performing breakout investments while neglecting struggling bets.
```mermaid
flowchart LR
accTitle: Scalable vs Traditional Support Comparison
subgraph Traditional["Traditional Venture Model"]
T1["Portfolio Growth"] --> T2["Staff Bottleneck"]
T2 --> T3["Triage Resources"]
T3 --> T4["Orphan Struggling Startups"]
end
subgraph SignalFire["SignalFire Software Model"]
S1["Portfolio Growth"] --> S2["Automated Data Platform"]
S2 --> S3["Universal Base Access"]
S3 --> S4["No Company Left Behind"]
end
```
Early-stage startups rarely follow a linear growth trajectory; they frequently pivot and search for product-market fit. Because SignalFire delivers platform support through automated software and data products, "no company gets left behind" or orphaned during transitional phases.
- **The Discord Precedent:** Discord initially launched as an early-stage gaming company and was effectively orphaned by its original seed investors during lean times. After successfully pivoting to its communication platform, the relationship was fractured, and the founders refused to accept follow-on capital from their original seed backer.
- **SignalFire's 100% Pro Rata Win Rate:** When portfolio startups break out and encounter intense multi-term-sheet competition at Series A and B, founders consistently prioritize investors who provided real utility. SignalFire maintains a **100% success rate** doubling and tripling down into its best-performing companies across competitive follow-on rounds.
---
[16:08](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=968s) Case Studies: Scaling Horizon3.ai and EvenUp
Real-world portfolio engagements highlight how SignalFire blends programmatic data access with targeted technical and executive guidance:
#### 1. Horizon3.ai (Autonomous Penetration Testing)
Horizon3.ai provides autonomous, AI-driven penetration testing by using software to impersonate state-level threat actors (such as Russian or North Korean state hackers). The tool acts as a sparring partner inside enterprise environments to expose, remediate, and continuously verify security vulnerabilities.
- **Founder Background:** Snehal Antani previously served as CTO of Splunk and later completed four years of government service as the inaugural CTO of the Joint Special Operations Cyber Command (JSOC).
- **Hiring Support:** Because the founder already possessed deep access to engineering talent exiting the national security and military sectors, the startup did not need IC hiring support. Instead, SignalFire focused on executive leadership, sourcing roughly **75% of the company's executive team** (including their CRO and Head of People) and directly participating in VP of Engineering technical interviews.
#### 2. EvenUp (AI Legal Intelligence)
EvenUp operates within the personal injury legal market, serving an industry of approximately 40,000 U.S. law firms. The platform uses AI to predict the settlement value and probability distribution of personal injury claims by digesting raw inputs such as police reports and medical records, enabling attorneys to benchmark case values accurately before entering settlement negotiations.
[19:14](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=1154s) EvenUp: Transforming Legal Workflows and Scaling with SignalFire’s Talent Engine
In personal injury law, determining the true market value of a settlement has traditionally been a guessing game. While any individual law firm only has visibility into its own narrow slice of cases, roughly 95% of personal injury cases settle out of court. Because settlement terms remain strictly private, no single firm has ever possessed a comprehensive view of the entire market.
Portfolio company EvenUp solved this structural information asymmetry by establishing what functions as an anonymous "data sharing co-op." Member law firms contribute their proprietary case settlement data into the platform and, in return, gain aggregated access to peer benchmarks to understand the real-time market value of any given case. Building on top of this unique dataset, EvenUp utilizes generative AI to draft comprehensive demand packages and reports detailing the facts of a case for judges or mediation panels—effectively automating the heavy labor historically handled by paralegals.
```mermaid
flowchart LR
accTitle: EvenUp Workflow Architecture
subgraph S1["Data Layer"]
A["Private Law Firm Data"] --> B["Anonymous Data Co-Op"]
B --> C["Market Valuation Benchmarks"]
end
subgraph S2["AI & Deliverable"]
C --> D["GenAI Analysis Engine"]
E["Case Facts"] --> D
D --> F["Automated Legal Reports & Mediation Packages"]
end
```
Despite having deep industry domain knowledge, EvenUp's founding team consisted of lawyers who originally lacked a technical co-founder. SignalFire stepped in to bridge this technical gap using its proprietary talent platform:
- **Key Technical Hires**: SignalFire sourced and placed one of the former heads of data science at Quora to serve as EvenUp's first Head of Data Science.
- **Full-Stack Recruiting Support**: SignalFire helped build out EvenUp's entire engineering and technical team from inception through their Series B round. For the first two years of the company's existence, all recruiting was executed entirely through SignalFire's internal tools.
- **Valuation Trajectory**: EvenUp subsequently reached a valuation of approximately $1.25 billion, followed by a subsequent financing round at a multiple of that figure.
---
[20:52](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=1252s) Delivering Scalable Value to Founders and Limited Partners
In closing, this operational approach highlights how modern venture capital is evolving beyond passive capital allocation. By building proprietary AI, data pipelines, and scalable software tools, venture firms can systematically solve core operational bottlenecks—like executive hiring, technical recruiting, and strategic intelligence—for their portfolio companies. This model creates a compounding flywheel that not only accelerates startup growth but also drives superior, repeatable returns for limited partners (LPs).
## Notable quotes
> [01:15](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=75s) if a VC firm were forced to pitch other venture capitalists to raise capital, it would fail the very diligence tests it applies to startups. — Ilya Kirnos
> [01:15](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=75s) apprentice industry — Ilya Kirnos
> [02:44](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=164s) one of one — Ilya Kirnos
> [02:44](https://www.youtube.com/watch?v=X0TcxgFvAnc&t=164s) one of many — Ilya Kirnos
## Video description
Ilya Kirnos, Co-Founder & CTO, SignalFire
Abstracts
Are you wondering how venture capital firms can truly differentiate themselves and provide lasting value in today's competitive landscape? The talk will explore why simply having capital or a data engineer isn't enough to stand out or become a genuine AI-driven fund. I will explain our strategy building a firm specifically designed with scalable software value-adds to ensure robust support for our growing portfolio companies. You will learn what founders prioritize beyond just funding and why operational excellence is becoming increasingly critical for VCs. Attend this session to understand how to build a future-proof venture model that genuinely serves founders and scales effectively.
SignalFire co-founder Ilya Kirnos argues that traditional venture capital operates with an ironic deficiency of technology, asserting that firms must engineer proprietary data pipelines, AI, and software to scale sourcing, diligence, and post-investment operational support.
Engineering First Principles and the Venture Flywheel (1:15)
SignalFire co-founder and CTO Ilya Kirnos, an engineer with backgrounds at Oracle and Google, founded the firm on the premise that venture capital was ripe for first-principles modernization. While allocating capital to cutting-edge technology companies, traditional VC firms historically operated via manual workflows, personal networks, and subjective intuition. Kirnos observed that if venture firms were forced to pitch other VCs to raise capital, they would fail the diligence tests applied to startups due to a lack of proprietary software or defensible product differentiation.
To address this gap, SignalFire—which manages approximately $3 billion in assets under management, including an active $1 billion vehicle focused on Seed through Series B—deconstructed early-stage investing into four core pillars: sourcing, diligence and selection, winning deals, and post-investment operational support. Kirnos maintains that delivering measurable post-investment utility triggers positive founder backchannel references, creating a self-reinforcing reputational flywheel that steadily increases win rates on competitive deals.
Data-Driven Sourcing with Human-in-the-Loop Diligence (5:25)
To identify breakout companies early, SignalFire built an enterprise-scale data infrastructure tracking roughly 80 million companies and more than 600 million professional profiles in real time. The platform ingests external datasets, resolves identity fragmentation, and applies machine learning algorithms to detect early momentum indicators, including engineering talent migration, revenue inflection points, web traffic surges, and mobile application download velocity.
Despite this level of automation, Kirnos stresses that early-stage venture capital remains fundamentally a human-in-the-loop discipline. While algorithms surface the top percentile of candidate startups and benchmark them against immediate peers, human partners must evaluate qualitative founder attributes that software cannot measure—such as vision, charisma, resilience, salesmanship, and the personal ability to recruit top-tier talent.
Software-Driven Support and Eliminating Startup Orphanage (12:50)
In traditional venture firms, operational support relies on human teams that degrade as the portfolio expands, forcing funds to triage resources toward breakout winners while effectively orphaning struggling or pivoting investments. Kirnos points to Discord's early experience as a cautionary precedent where initial seed investors neglected the company prior to its pivot, permanently damaging relationship equity. SignalFire claims its automated, software-delivered support provides universal baseline utility, enabling the firm to report a 100% success rate in winning competitive pro rata allocations across follow-on rounds.
SignalFire structures its operational support to bifurcate individual contributor hiring into self-serve software and executive search into a white-glove in-house team. The firm also repurposes its sourcing database into go-to-market prospecting tools for founders. Kirnos notes that portfolio needs shift alongside macro cycles: during the capital-abundant 2021 peak, support focused heavily on engineering recruiting, whereas the higher-interest-rate regime shifted priority toward capital efficiency, tech-stack mapping, and targeted GTM conversion.
Portfolio Deployments: Horizon3.ai and EvenUp (16:08)
Real-world engagements illustrate how programmatic data access is paired with targeted operational assistance. For Horizon3.ai, an autonomous penetration-testing startup founded by former Splunk and JSOC Cyber CTO Snehal Antani, the company already possessed deep access to technical talent exiting the national security sector. SignalFire bypassed individual contributor hiring to focus on leadership, sourcing approximately 75% of the startup's executive suite, including the Chief Revenue Officer and Head of People.
In the legal technology vertical, EvenUp addressed the lack of visibility into private personal injury settlements by building an anonymous data-sharing cooperative among law firms, pairing proprietary data with generative AI to draft mediation reports. Because EvenUp's lawyer founders initially lacked a technical co-founder, SignalFire sourced the company's first Head of Data Science from Quora and powered its engineering hiring through internal tools through Series B, supporting the startup's ascent to a valuation of approximately $1.25 billion.
Kirnos concludes that replacing artisanal venture practices with proprietary software and data platforms creates compounding advantages for founders and limited partners alike. However, the model requires continuous calibration between automated signal detection and qualitative human diligence, particularly as market cycles dynamically alter what portfolio founders require to survive.
[1:15] Ilya Kirnos: if a VC firm were forced to pitch other venture capitalists to raise capital, it would fail the very diligence tests it applies to startups.
[1:15] Ilya Kirnos: apprentice industry
[2:44] Ilya Kirnos: one of one
[2:44] Ilya Kirnos: one of many
[2:44] Ilya Kirnos: calling card
[5:25] Ilya Kirnos: human-in-the-loop
[9:35] Ilya Kirnos: machine for getting talent in the door
[9:35] Ilya Kirnos: ZoomInfo on steroids
[12:50] Ilya Kirnos: no company gets left behind
[19:14] Ilya Kirnos: data sharing co-op
Original transcript
[0:04] Hi everyone. Uh good to be here. Uh
[0:06] before I dive into the presentation,
[0:08] quick bio on me. Um software engineer by
[0:10] background. CSMath undergrad. Came out
[0:13] to Silicon Valley right after college.
[0:15] Um I went to college in the east coast
[0:17] 25 years ago now. Uh began my career at
[0:20] Oracle in the database team and then
[0:22] after a few years at Oracle, moved over
[0:24] to Google in 2003. Uh at Google I worked
[0:26] on uh several things. I was there for 9
[0:29] years uh AdWords uh infrastructure then
[0:32] Gmail Gmail monetization left Google in
[0:35] mid 2012 to join a startup that was
[0:37] acquired by Twitter pretty quickly and I
[0:39] left before the acquisition and
[0:41] co-founded SignalFire uh and really my
[0:43] path into venture capital uh was as a
[0:46] technical co uh co-founder and CTO. Uh
[0:49] SignalFire is unique in that we are the
[0:52] first venture capital firm built like a
[0:53] tech company. Uh currently we have about
[0:56] three billion under management. uh
[0:58] current fund is about a billion. We
[1:00] invest across stages uh starting with
[1:02] seed all the way to probably I would say
[1:04] series C is where we uh kind of tap out
[1:07] but really the core of the funds is
[1:09] focused on series seed through A and B.
[1:12] Um and um uh let me advance the slide
[1:15] here. One interesting thing to me about
[1:16] the VC industry is um you know there's
[1:19] an irony in what we do in that we invest
[1:22] in these cutting edge technology
[1:23] companies and yet our own industry
[1:25] itself is quite tech deficient. If you
[1:28] look at most venture capital firms
[1:29] especially when we started about 12
[1:31] years ago um most of them used zero
[1:34] technology. it was all manual services
[1:37] and um you know I always kind of did
[1:39] this thought exercise of if I were a
[1:42] venture capital firm and I went to pitch
[1:43] other VCs to raise capital what would
[1:46] that look like you know all the same
[1:47] questions we ask our uh startups when
[1:50] they pitch us I think most VC uh firms
[1:53] don't have great answers to uh what
[1:55] technology you use well none it's all
[1:57] manual what's your product
[1:59] differentiation not really any I mean
[2:01] there's reputation and brand is probably
[2:04] the biggest on um and so we thought that
[2:07] there there was room to use data and
[2:10] software and apply it to venture capital
[2:12] in the same way that it was being
[2:13] applied to basically every other
[2:16] vertical, every other industry. Um and
[2:19] uh there was a lot of skepticism around
[2:21] this about 12 years ago when we started.
[2:23] I think the perception was venture
[2:24] capital is very much an apprentice
[2:26] industry. It's an art. It's not a
[2:28] science. And how would you use data uh
[2:31] to um to uh how would you apply data to
[2:34] early stage venture? Because you know
[2:36] what is even the data at those early
[2:38] stages when you have a sestage company
[2:39] with nothing but a team and an idea. And
[2:41] so we approach this very much uh from
[2:44] first principles. And when you think
[2:46] about venture capital, you can really
[2:48] break it down into what I would say are
[2:50] four pillars of success. You know, you
[2:53] need to do four things well to win in
[2:55] venture capital. The first one is
[2:57] sourcing, right? You need to be able to
[2:59] identify companies as they're getting
[3:01] started and ideally identify them and be
[3:04] the first investor to talk to them
[3:06] because that gives you an edge uh in
[3:08] investing in them. Uh the second piece
[3:10] is diligence and selection. Right? Let's
[3:13] say you can identify interesting
[3:14] companies. The next question is okay is
[3:16] this specific company one of one in its
[3:18] space or is it one of many? All right.
[3:21] And if it is one of many uh how does it
[3:23] stack up against its competitors? So
[3:25] imagine back in the day you're talking
[3:26] to Uber, you're talking to Lyft, and
[3:28] you're talking to the eight other
[3:30] companies in the ride sharing space. How
[3:32] do you ensure that you pick ideally Uber
[3:35] or maybe Lyft uh but not the others,
[3:37] right? Because the others nobody even
[3:39] knows their names anymore. Um so that's
[3:41] kind of the second pillar. The uh third
[3:44] pillar here um uh is postinvestment
[3:47] support, but I would actually insert an
[3:49] intermediate stage which is winning.
[3:51] Right? If you see a company and you want
[3:53] to invest in the company, most startups
[3:56] um that are doing well have their option
[4:01] uh options as far as uh which investors
[4:03] to take money from. And um you really
[4:06] need a sharp calling card to be able to
[4:08] win um and and kind of be the investor
[4:11] of choice for founders. And that leads
[4:13] me to number three here, which is post
[4:15] investment support. Obviously, once the
[4:18] company's inside your portfolio, you
[4:20] want to be able to uh support them in
[4:24] their growth as much as you can as a
[4:25] venture capitalist because it benefits
[4:27] you directly. But there's a secondary
[4:29] benefit to supporting your portfolio
[4:31] companies super well and that is winning
[4:34] in the future, right? And the reason is
[4:36] that when founders are doing reference
[4:39] references on you as an investor, one of
[4:41] the top things they do is talk to your
[4:43] existing portfolio founders. You know,
[4:44] founders talk to founders all the time.
[4:47] And if your existing portfolio founders
[4:49] say great things about you and say, you
[4:51] know, Signal Fire promised us all this
[4:54] stuff and they've really been able to
[4:55] deliver, that makes it that much easier
[4:57] for you to win in the future when you're
[5:00] competing for that next investment,
[5:01] right? And it really starts your
[5:03] reputational flywheel spinning as a
[5:05] venture capital firm because it makes it
[5:07] easier for you to win that next deal.
[5:09] And then what when that founder says
[5:11] great things about you to other founders
[5:14] then you can win the next deal the next
[5:16] one the next one etc. And so that's how
[5:18] we think about uh kind of breaking down
[5:20] venture capital into kind of first
[5:22] principles if you will and applying data
[5:25] to each one. And so when it comes to
[5:27] sourcing um we are tracking something
[5:30] like 80 million companies in real time
[5:33] um and 600 uh over 600 million
[5:35] individual profiles of people um and
[5:38] lots of signals lots of data points on
[5:40] each one and from a technical
[5:41] perspective you can think of us as
[5:43] taking data uh that exists in disparate
[5:46] silos bringing it all under one roof uh
[5:50] organizing it cleansing it kind of like
[5:52] the you know the classic data
[5:54] engineering problems that uh are
[5:56] encountered in dealing with lots of data
[5:59] and then we um run ML AI models on top
[6:03] to help us identify the fastest growing
[6:06] companies, the most talented uh
[6:08] individuals and given domains and
[6:10] different uh different functional roles
[6:12] etc. And that all comes together for
[6:14] example for early identification, right?
[6:16] And so you can basically say which
[6:18] companies are attracting the best
[6:20] talent, which companies are growing
[6:22] revenue the fastest, which companies are
[6:24] seeing a surge in uh traffic to their
[6:27] website or in surge in downloads from
[6:29] their um in the app store for their
[6:31] apps, right? And those uh companies then
[6:33] bubble up and then they're handed over
[6:36] to our human investment team to kind of
[6:38] take it to the next step in the funnel.
[6:39] Not all of this can be automated. You
[6:41] know, we believe that what wins in
[6:42] venture capital is the best of data
[6:45] intelligence and the best in human
[6:47] intelligence because there's no really
[6:49] substitute for talking to a founder uh
[6:52] understanding their vision, their
[6:53] excitement, their passion, their ability
[6:55] to sell, raise capital, recruit, etc.
[6:58] Right? So, it's very much a human in the
[7:00] loop problem. And so, once you identify
[7:02] a company, you can then um take a
[7:04] datadriven decision on whether to take
[7:06] it to the next stage in the funnel.
[7:08] Right? That's where the humans come in.
[7:09] and also gives you a lot of
[7:10] benchmarking, a lot of context because
[7:13] as I mentioned, you really want to know
[7:14] as an investor, is this company one of
[7:16] one or one of many. Um, and our data
[7:19] allows us to basically say, you know,
[7:20] for this given company, here are the
[7:22] five other companies that are relevant
[7:25] that are in that same space and here's
[7:27] how this company benchmarks against
[7:29] those other five on the dimensions that
[7:31] we care about, right? And then of
[7:33] course, post investment support comes in
[7:36] uh using the same data. You know, if you
[7:38] look at sort of our lens on this as an
[7:41] investor, uh, when we're doing this
[7:43] benchmarking, once the company is inside
[7:45] our portfolio, founders like to get
[7:48] competitive intelligence on their peers
[7:50] in the space, right? So, the same data
[7:52] is really interesting to them for
[7:54] understanding how they're uh competing
[7:56] out there in the market versus their
[7:57] peers. Another way you can use this um
[8:00] the same data that we have is for uh
[8:03] recruiting. It's a huge pain point for
[8:05] all of our portfolio companies.
[8:07] Attracting the best talent, especially
[8:08] technical talent, engineering talent is
[8:10] a huge pain point. And then go to market
[8:12] is another huge pain point that we've
[8:14] seen. And you know, these things vary
[8:16] with market cycles. If you think back, I
[8:18] don't know 2021, you know, the uh as we
[8:22] kind of in retrospect now know kind of
[8:24] like the peak of the covid bubble, you
[8:25] know, the the number one pain point in
[8:27] our portfolio was attracting great
[8:29] talent. Companies had plenty of capital.
[8:32] uh customers were spending money freely.
[8:34] You know, sales go to market was not the
[8:36] biggest concern. The biggest um uh uh
[8:38] bottleneck to growth was getting enough
[8:40] people in the door, getting talent in
[8:42] the door. So, there was a lot of demand
[8:44] for talent and recruiting. Now, fast
[8:45] forward just a couple of years was
[8:47] rising interest rates. What happened
[8:49] was, you know, companies started having
[8:51] layoffs. You know, Google had layoffs,
[8:52] Facebook laid people off, talent became
[8:55] more plentiful. All of a sudden the
[8:56] needs of our portfolio shifted and it
[8:59] became much more um important to be
[9:01] growing fast but efficiently. It was no
[9:03] longer growth at all costs. And so to
[9:05] the extent that we could help our
[9:07] companies grow more efficiently uh using
[9:09] our data that became super important.
[9:11] And so we have offerings to our
[9:13] portfolio and this is by the way this is
[9:15] all free of charge. This is how we
[9:17] compete and win. Um you know we say if
[9:19] you go with signal fire you get access
[9:21] to all this data to all these tools. So
[9:23] we provided go to market support. For
[9:24] example, you can uh build a prospecting
[9:27] list using our data and go really
[9:29] deeply. Let's say you are a uh data uh
[9:32] data platform company, right? You want
[9:34] to understand which companies are
[9:35] growing their data science teams the
[9:37] fastest. We can give you uh that
[9:39] information. We have that data, right?
[9:41] And so this is how we think about
[9:43] applying data and software to venture
[9:45] capital. And one of the things that I
[9:47] want to delve deeper into is and I
[9:49] started talking about this a little bit
[9:50] is this um post-investment support. And
[9:53] you can see that you know there are
[9:55] really a a a few key areas where all of
[9:58] our companies have um tremendous needs.
[10:00] You know for talent acquisition you can
[10:02] think of that as falling into two
[10:04] buckets. One is what I would call volume
[10:06] recruiting. You know individual
[10:08] contributors where you're constantly
[10:09] having to recruit. And in some ways the
[10:12] bigger you get the more of a pain point
[10:14] this becomes when you're a sestage
[10:16] company when you have just the founders
[10:18] and maybe the initial engineers uh you
[10:20] know getting one or two engineers more
[10:22] doubles the size of your engineering
[10:23] team you don't need to hire an army as
[10:25] you get to the series B series C growth
[10:28] stage you need to be creating you know a
[10:30] machine for getting talent in the door
[10:32] because it's a continuous need um then
[10:34] there's executive recruiting right and
[10:37] so I I'd say that's more of a white
[10:38] glove service where you're not always
[10:41] having to hire a new CRO or a new CMO.
[10:44] That's a much more point in time need.
[10:46] And for that, we actually have a
[10:48] separate team that does more white glove
[10:50] services. So, the volume uh recruiting
[10:52] goes through our products. It's more
[10:54] self-s serve because it's a constant
[10:56] need. Um and it's uh it's something that
[10:59] you you need to be able to scale. And so
[11:01] products make sense there. For something
[11:03] that's more white glove, we have a team
[11:05] that comes from, you know, the executive
[11:08] search team at Facebook and Google and
[11:11] Whimo, etc. that helps our companies
[11:13] think think through hiring their next uh
[11:15] executive talent. Uh go to market. I
[11:18] mentioned uh we look at uh think of the
[11:21] data that we have as a little bit maybe
[11:23] like uh Zoom info on steroids, right? We
[11:25] are very specialized to early stage
[11:28] startups in technical areas. And so we
[11:31] can give them data and prospecting lists
[11:33] and insights that they can't get out of
[11:35] a generic solution out there in the
[11:37] market. And we think about that all the
[11:38] time because we don't want to go out and
[11:40] compete with the myriad of vendors, some
[11:44] of whom are at this conference. We want
[11:46] to really focus on the things that we're
[11:47] best at and that's where we really uh
[11:49] kind of pinpoint our efforts. And then
[11:53] uh I mentioned competitive intelligence.
[11:55] You know if you were we were investors
[11:57] in Uber, right? If you think about who
[12:00] Uber was competing with day-to-day on
[12:02] the ride sharing side, it was Lyft
[12:04] companies like Lyft, right? And that
[12:06] competition took the form of kind of
[12:09] every market was its own entity, right?
[12:12] Lyft in LA was very different from Lyft
[12:14] in New York. You know, LA was a huge
[12:16] area of strength for for Lyft. They
[12:18] weren't as as prevalent in New York, for
[12:20] example. And so using our data, we were
[12:22] able to slice and dice kind of the the
[12:25] revenue and the retention and the
[12:27] customer uh lifetime value of um of a
[12:31] rider by all these cities and share it
[12:34] with Uber, our portfolio company, to
[12:36] enable them to grow more efficiently and
[12:38] figure out how to spend their marketing
[12:40] dollars in these different geos. Um and
[12:43] so um I know we're kind of uh running up
[12:46] on time, so let me kind of keep going
[12:48] here. Uh so one of the advantages that
[12:50] you have here is with traditional VC
[12:52] because everything is manual in some
[12:54] ways the more portfolio companies a
[12:57] venture capital firm gets the worse the
[13:00] service is for any one company right if
[13:03] you have let's say 10 recruiters on
[13:05] staff that works great when you have 10
[13:07] portfolio companies because you can
[13:08] really dedicate time and uh dive deep
[13:11] into each one. when your portfolio grows
[13:14] to 100 companies, all of a sudden that
[13:16] breaks everything, right? And what
[13:18] happens in most traditional venture
[13:20] capital firms is you begin to triage.
[13:22] You say, "Okay, this company is doing
[13:25] really well. It's going to move the
[13:26] needle on our portfolio. It's going to
[13:28] be a driver of returns for us. Let's
[13:30] focus all our efforts on that." And that
[13:32] makes sense. You know, rationally, it
[13:34] makes sense for traditional venture
[13:36] firms to do that. However, what um we've
[13:39] seen in our model is when you are able
[13:41] to deliver support that scales uh
[13:44] through products, through data, through
[13:46] software, it means that no company gets
[13:49] left behind. And one interesting thing
[13:51] about the early stage of uh stages of
[13:53] investing is uh companies pivot a lot.
[13:58] You know, it's rare that a company sort
[14:00] of goes in a straight line all the way
[14:02] up. A lot of companies take some time to
[14:04] get to product market fit, maybe pivot,
[14:06] etc. And what this means is no companies
[14:09] get orphaned. You know, in a traditional
[14:10] venture firm model because you're
[14:12] focused on the companies that are
[14:13] performing the best, you tend to orphan
[14:16] the companies that are struggling,
[14:17] right? And those companies that are
[14:18] struggling today may be uh uh the next
[14:22] stars of tomorrow. And there are all
[14:23] kinds of examples of companies like
[14:25] that. For example, you know, if you look
[14:27] at Discord, uh started out as a gaming
[14:29] company, right? was orphaned by their
[14:32] original investors. Then they pivoted
[14:34] into chat uh kind of what they're known
[14:37] for today and they that relationship was
[14:40] fractured. You know, they never took
[14:41] money from that seed investor again
[14:43] because they felt like they'd been
[14:44] abandoned when tough uh when times were
[14:46] tough initially. That doesn't happen in
[14:48] our model because every company feels
[14:50] the love from Signal Fire because it's
[14:52] scalable. our services are scalable and
[14:55] you know we don't need to perform
[14:56] individual heroics uh kind of manually
[14:58] to be able to support our portfolio.
[15:02] Uh as a result uh we get really high
[15:05] marks from our portfolio. you know, we
[15:06] do these surveys and try to understand,
[15:08] ask them who's the most um valuable
[15:11] investor on your cap table and we get
[15:13] really high marks and one of the uh kind
[15:16] of best ways that we track this is when
[15:19] companies perform uh what happens is
[15:21] they get a lot of investor interest. You
[15:23] know, we've had any number of companies
[15:24] that have broken out and on their you
[15:27] know, when we lead the seed round and
[15:29] they're raising their series A or their
[15:30] series B, you know, they get 10 term
[15:32] sheets, right? and we would love to
[15:34] deploy more capital into the company
[15:36] when it's performing. Uh you have all
[15:38] this competition. Well, what happens is
[15:40] if you've delivered for the company,
[15:41] they will gladly take your money and if
[15:43] you haven't, if your promises have gone
[15:45] unfulfilled, they probably will go with
[15:47] a different investor. And uh we have
[15:50] never failed as a firm, 100% success
[15:52] rate, doubling and tripling down into
[15:54] our best performing companies in the
[15:56] most uh competitive rounds. Um and that
[15:59] really to me speaks to the our ability
[16:02] to deliver value to these companies
[16:04] through data software at scale. Um and
[16:08] uh maybe a couple of uh in in closing a
[16:10] couple of specific examples of companies
[16:12] where we've been able to sort of uh
[16:14] supercharge their growth. Uh the first
[16:17] one I want to talk about talk about is
[16:18] Horizon 3.ai. This company you can think
[16:21] of it as an AI hacker. It's an AIdriven
[16:24] penetration testing company. And what
[16:25] they do is basically using software and
[16:28] um and AI, they impersonate a state
[16:32] level attacker, right? And so think of
[16:34] them as a sparring partner for your
[16:36] security posture. You know, you let them
[16:38] loose in your environment. They go out,
[16:40] find weaknesses much like a uh Russian
[16:43] or a North Korean hacker might, and then
[16:46] come back to you and report on the gaps
[16:47] they have found, allowing you to fix
[16:50] those gaps and then verify that they've
[16:52] been fixed and they stay fixed. Right?
[16:54] So this this company is interesting. The
[16:56] founder of this company was previously
[16:58] the CTO of Splunk. Then he went into uh
[17:02] government service for about four years
[17:04] and was the first CTO of JOCK which is
[17:07] the joint special operations cyber
[17:09] command. Um so he had a deep network in
[17:11] cyber security had a lot of access to
[17:14] talent from the national security
[17:16] community. As they were leaving the
[17:18] military he was able to get them into
[17:21] Horizon 3. So in his case, he didn't
[17:23] necessarily have a need for uh
[17:26] individual contributor recruiting in
[17:28] those initial stages because he had a
[17:29] great network and access to talent. But
[17:32] what he did need as the company evolved
[17:33] was executive recruiting. Right? As the
[17:36] company scaled up, he was no longer able
[17:38] to do all the selling himself. We needed
[17:40] to recruit a CRO. Um as the company
[17:43] scaled up, he no longer was able to take
[17:45] care of all the HR issues himself. We
[17:47] need to find a a head of people, right?
[17:49] And so, uh, if you look at his executive
[17:52] team today, the company's executive team
[17:54] today, uh, you're looking at, you know,
[17:57] a team that was probably 75%,
[18:01] uh, recruited through signal fires help.
[18:04] And to this day, I still help interview.
[18:06] So, right now, we're uh, looking for VP
[18:08] engineering, for example, and I'm sort
[18:10] of in the interviews doing those
[18:11] interviews for the company. Now, in the
[18:13] case of Evenup, uh, the situation was a
[18:15] little bit different. So, Evenup is in
[18:17] the legal space. They sell to personal
[18:19] injury law firms. I'm sure many of you
[18:22] as you took your Ubers or taxis from the
[18:25] airport, saw those billboards up uh on
[18:27] on the highway, you know, injured, call
[18:30] 1800, better call Solid or whatever, you
[18:32] know, the these injury law firms. Turns
[18:34] out it's a huge industry in the United
[18:36] States, about 40,000 law firms,
[18:38] something like that. And what this
[18:39] company does is it allows these law
[18:42] firms to put a predicted value on a
[18:46] case, right? So when you walk in the
[18:48] door and you say, you know, I was riding
[18:50] my bike and this uh, you know, bad guy
[18:52] Ilia hit me with his car and now, you
[18:54] know, I have pain and suffering. He owes
[18:56] me a billion dollars. Right? Most of
[18:58] these cases are settled. And looking at
[19:00] the facts, kind of the police report,
[19:01] the medical medical report, the in
[19:04] personal injury firm will put a
[19:07] probability on the likelihood that
[19:09] you'll settle and not go to trial and
[19:12] also what the likely amount of the
[19:14] settlement will be. Well, any one
[19:16] personal injury law firm has their
[19:19] slice, their view of the market, but
[19:21] they don't know the overall market. And
[19:23] because this data is private, 95% of
[19:26] cases settle out of court, no one has a
[19:28] great view on the to totality of cases.
[19:31] So what evenup has started uh has
[19:33] enabled is effectively think of it as a
[19:35] data sharing co-op where law firms they
[19:38] contribute anonymously their data and
[19:40] they're able to access their peers data
[19:43] to understand the value current current
[19:45] market value of a case and then uh this
[19:48] company uses Genai to actually generate
[19:50] the report that goes in front of a
[19:52] mediation panel or a judge uh with the
[19:55] facts of the case and really replaces
[19:57] the labor of what parallegals would have
[19:59] to do in the
[20:00] Um, and so the company has been growing
[20:02] super fast, but what was interesting
[20:03] about it is the company didn't really
[20:06] have a technical co-founder when they
[20:07] started. Uh, it was a bunch of lawyers
[20:10] kind of coming out of the industry. So,
[20:11] they knew the domain super well, but
[20:13] what they needed was help on um on the
[20:15] technical side and talent uh and help on
[20:17] recruiting their technical talent. And
[20:19] so one of the first things we did was we
[20:21] recruited um uh one of the heads of data
[20:24] science of Quora as the first head of
[20:26] data science for this company and then
[20:27] helped them recruit their entire
[20:29] technical team all the way through their
[20:30] series B. You know this is a company
[20:32] that recently or maybe uh 18 months ago
[20:35] raised about uh a billion and a quarter
[20:38] valuation. There's another round that
[20:40] has not been announced that's been
[20:41] raised that's a multiple of that. So
[20:43] this company is doing very well. And so
[20:45] the you know for the first two years of
[20:46] their existence they were doing all the
[20:48] recruiting through the tools that we
[20:50] provided to them. Right? So um I know
[20:52] I'm up on time super quickly. Uh thank
[20:54] you for your time. Uh hopefully that
[20:56] gives you a little bit of insight into
[20:57] how venture capital works um and how
[21:00] we're able to leverage AI data software
[21:02] to uh really deliver value to our
[21:05] companies and uh deliver value to our
[21:07] limited partners, our investors as well
[21:09] and provide great returns. Thank you so
[21:10] much.
