In addition to helping venture capitalists decide which startups to fund, AI is also helping us figure out which to avoid.
VCs face a challenging asymmetry of information: startups really know what’s happening inside a business while investors do their best to see through any smoke or mirrors. This is especially difficult when charismatic founders are surrounded by all the right people, investors, buzzwords and narratives. While it may still take a human to detect certain forms of bullsh*t (BS), it turns out that AI can be remarkably effective at spotting other forms of BS that can elude even the savviest investors.
That was the case in this week's scandal involving a startup called 11x.ai.
The startup, positioned as an AI-powered sales automation platform, had raised eyebrows for its meteoric rise. In just two years, it reported approaching $10 million in annual recurring revenue (ARR) and successfully raised a $50 million Series B round. The round valued the company at $350 million post-money, drawing investment from some of the most respected names in venture capital, including Andreessen Horowitz, Benchmark, 20VC, Project A Ventures, Lux Capital and SV Angel. [1] All seemed to be going well for 11x.ai, with even Pitchbook's "VC Exit Predictor" giving the company a 75% probability of success.
Despite the sophistication of 11x.ai's investors, they were misled. According to TechCrunch and multiple insiders, 11x.ai appears to have exaggerated both its revenue and its customer base. Several of the marquee clients listed on its website (such as ZoomInfo and Airtable) were never actually customers beyond short-term pilots, and their logos were used without consent. [2,3]
More significantly, while the company claimed nearly $10 million in ARR, the actual revenue retained beyond short-term trial contracts was closer to $3 million.[4] This gap was the result of internal accounting that treated temporary trial users as if they were full-year contracted customers, effectively inflating the company’s revenue narrative and giving the illusion of greater traction. [5]
Detecting BS is a significant challenge when investing in startups, and private markets overall. Startups, by their nature, operate in opaque environments. There can be little to no public disclosure, SEC filings or other reliable data. Founders also have every incentive to tell a compelling story, and in some cases, that story can drift too far from the truth. Even the best investors, armed with diligence processes and industry insight, can be caught off guard when relying on data that's incomplete or misleading.
BS detection is where AI can make a difference. Not in the future, but now.
For example, Quannix®, the AI platform we use to analyze startups, raised red flags about 11x.ai and its valuation long before the scandal surfaced. At the time of the $350 million valuation, Quannix® estimated 11x.ai’s value at just $31 million. In other words, AI took the position that the startup - backed by over $75 million from some of the biggest names in venture capital - was overvalued by roughly 11X. In fact, Quannix® estimated that the startup wasn’t even worth as much as the total capital it had raised. Not for nothing, that's gutsy.
In Q1 of this year (but before the public allegations) the AI's estimate had grown, yet still remained below $45 million.[6] The gap between what the AI thought 11x.ai was worth, and what investors thought it was worth, was large, persistent and difficult to ignore.
While Quannix® didn’t have access to 11x.ai’s internal revenue data, if the company’s actual ARR was indeed around $3 million, as now appears likely, the AI valued the startup at roughly 10 times that figure. That is a generous, but still reasonable, revenue multiple for a fast-growing AI startup. That being the case, it was far more grounded than the 35 times revenue multiple that Series B investors believed they were paying. In reality, once the scandal arose, they realized they'd paid closer to 116 times revenue. Whatever the yardstick, when the dust settled the AI was a lot more accurate in revealing 11x.ai’s true value and traction.
Quannix® didn't rely on pitch decks, founder presentations or board updates. It had no access to confidential metrics, revenue details or internal projections. Instead, it looked at a broad array of data observable from the outside, mined and aggregated purely by the AI itself (while ignoring funds raised or valuations from past or present funding rounds). According to the AI, on its independent basis, the company’s fundamentals simply didn't support the valuation it had received - not even close.
This is not an isolated case.
We saw a similar pattern with Thrasio, the Amazon aggregator once valued at $10 billion. Quannix® consistently estimated the company’s value as being nearly thirty times lower. When Thrasio ultimately filed for bankruptcy, it wasn’t a surprise to those following the AI’s signals.[7]
The same happened with IRL (short for 'In Real Life'), a social app that raised $200 million at a unicorn valuation, only for it to shut down after admitting that 95% of its users were fake. Quannix® had long valued IRL at less than one-tenth the level of its market valuation rather than being dazzled by the founder's false claims and storytelling. [8]
In each of these cases, AI provided something rare: a dispassionate, independent perspective that ultimately proved more accurate than the views held by even the most sophisticated investors and insiders (some of whom were misled by incomplete or fraudulent information).
It’s important to say this clearly: AI is not perfect. Still, it doesn't get caught up in excitement or charisma. It doesn't respond to herd behavior. It doesn't care how cool or persuasive the founders are, or how impressive the investors may be.
Instead, it offers an objective, data-driven external opinion about which deals to chase and which to run away from. That clarity is becoming increasingly valuable in a world where the gap between truth and narrative can be dangerously wide. The lesson from 11x.ai isn't that investors were foolish or careless. Rather, the lesson is that even the best can be misled when the inputs they are relying on are distorted. AI is helping to close this gap.
We are only beginning to understand the role artificial intelligence will play in venture capital, but it's already helping investors make more grounded decisions—seeing through illusions, valuing companies more accurately, and in some cases, catching warning signs long before they become headlines.
The 11x.ai story is a reminder that even in a market fueled by optimism, it’s helpful to have an AI around to quietly help call out BS when it matters most.
Endnotes
[^1]: TechCrunch. “A16z and Benchmark-backed 11x has been claiming customers it doesn’t have.” TechCrunch, March 24, 2025. https://techcrunch.com/2025/03/24/a16z-and-benchmark-backed-11x-has-been-claiming-customers-it-doesnt-have
[^2]: Ibid. ZoomInfo confirmed that it was never a customer and had demanded its logo be removed from 11x.ai’s website.
[^3]: Ibid. Airtable also denied being a customer, stating the product was never used in production.
[^4]: Pivot to AI. “11x.ai Juiced Its ARR, Got Sued, and Now the AI Community is Embarrassed.” Pivot to AI, March 25, 2025.
[^5]: Ibid. 11x.ai reportedly used “Contracted ARR” that counted three-month trial customers as if they were on annual plans.
[^6]: Internal Quannix model data, Q4 2024 and Q1 2025 estimates.
[^7]: Thurston, Thomas. “AI Dodged a $16B Bubble: Humans Weren’t So Lucky.” November 29, 2023. https://www.gsventures.com/news/ai-dodged-16b-bubble-humans-not-as-lucky
[^8]: Thurston, Thomas. “IRL App Faked 95% of Its Users: Humans Bought In, But AI Didn’t.” June 29, 2023. https://www.gsventures.com/news/irl-app-faked-95-of-its-users-humans-bought-in-but-ai-didn-t