- One Cerebral
- Posts
- Vertical AI Agents Could Be 10X Bigger Than SaaS
Vertical AI Agents Could Be 10X Bigger Than SaaS
Lightcone Podcast: Garry Tan, Diana Hu, Harj Taggar, and Jared Friedman
Jiten Patel
March 07, 2025
Credit and Thanks:
Based on insights from Y Combinator.Key Learnings
AI startups should replace entire workflows, not just improve existing tools—this is where billion-dollar opportunities lie.
Big incumbents fail in niche markets because they lack the patience for deep specialization—startups can win by mastering complexity.
The next unicorns will be AI-first companies with tiny teams, scaling through automation instead of headcount.
Selling AI automation requires targeting executives, not mid-level employees who fear being replaced.
The fastest-growing AI startups find their ideas by observing boring, repetitive tasks in industries they know well.
Today’s Podcast Host: Y Combinator
Title
Vertical AI Agents Could Be 10X Bigger Than SaaS
Speakers
Garry Tan, Diana Hu, Harj Taggar & Jared Friedman
Speaker Credentials
Garry Tan is the CEO of Y Combinator and co-founder of Initialized Capital, with a background as an early employee at Palantir Technologies and co-founder of Posterous. He holds a BS in Computer Systems Engineering from Stanford University and has been recognized on the Forbes Midas List from 2018-2022, with an estimated net worth of around $2 billion as of 2024, largely due to successful investments in companies like Coinbase and Instacart.
Harj Taggar is Group Partner at YC, who also a co-founder of Initialized Capital, played a key role in raising $39 million for the fund in 2013.
Jared Friedman is a Group Partner at YC, having co-founded Scribd.com in 2005 and served as its CTO, significantly contributing to its growth into one of the top 100 websites globally.
Diana Hu is a Group Partner at YC, and was co-founder and CTO or Escher Reality which was acquired by Niantic.
Podcast Duration
42:12
Read Time
Approx. 5 mins
Deep Dive
Jared makes a compelling case that vertical AI agents will be as transformative as SaaS but on an even greater scale. He points out that SaaS itself emerged from a technological inflection point—XML HTTP requests in the mid-2000s—which enabled rich web applications. Similarly, the rise of LLMs and AI automation is now enabling the next wave of software companies that don’t just build tools for humans but replace human workflows entirely.
One of the strongest arguments for vertical AI's growth is the structure of enterprise software itself. Unlike consumer tech, where dominant players like Google and Meta own entire categories, B2B SaaS has always been fragmented. There was never a “Microsoft of SaaS” because each vertical required deep domain expertise—handling niche regulations, workflows, and customer needs. That same fragmentation applies to AI automation: there won’t be one AI agent to rule them all, but rather hundreds of verticalized AI companies, each replacing a different specialized function.
The discussion highlights why big incumbents failed to build B2B SaaS startups and why history will repeat itself with AI. Giants like Google or Microsoft don’t have the patience to dive into highly specific, regulatory-heavy areas like payroll, medical billing, or compliance. Startups, however, can thrive in these niches by leveraging AI to replace entire operational teams. This is why early-stage AI startups are already getting rapid enterprise traction.
An example of this shift is the rise of AI-powered customer support agents. Despite a crowded market, most early entrants were simple LLM wrappers that couldn’t truly replace human workflows. Now, the new wave of startups is developing AI agents that don’t just assist human teams but fully replace them. A company called Outset, for example, is building AI-powered user interviews to conduct qualitative research. Similarly, Apriora is handling both technical and recruiter screening, removing the friction of selling to recruiters who might fear losing their jobs.
Hiring trends are also changing. Instead of aggressively building large teams, AI-native startups are hiring more engineers who specialize in LLMs to automate work that traditionally required large customer success, sales, and operations teams. The idea of a unicorn company with just ten employees is no longer far-fetched. This shift will fundamentally alter the economics of startups, reducing payroll costs while increasing profitability and efficiency.
For founders looking to identify the right vertical, the best opportunities lie in boring, repetitive admin work. Many of the most promising AI startups originated from founders simply observing inefficient manual processes—whether it was a friend manually refreshing government contract listings or a mother spending hours processing dental insurance claims. The key is to find high-cost, high-volume tasks that enterprises are willing to pay to automate.
AI voice calling is another explosive category. Startups like Salient are building AI-driven debt collection agents, automating an entire industry plagued by high churn and inefficiency. Similarly, customer service AI platforms are now handling tens of thousands of tickets daily, eliminating massive support teams. The rapid evolution of AI-generated voice technology—once considered years away—is now happening in real time, making AI voice startups some of the fastest-scaling businesses in the space.
The conversation also touches on the potential future of management. AI tools will enable CEOs to scale their impact far beyond traditional limits. A company recently built an AI voice agent that called all 1,500 of its employees, collected qualitative feedback, and summarized insights for leadership—something previously impossible at scale. This hints at a future where AI dramatically expands the number of employees a single leader can effectively manage.
Jared’s ultimate prediction: Just as SaaS created 300+ unicorns, vertical AI will do the same—but on an even larger scale. Startups that identify the right inefficiencies and build AI-native replacements will not only disrupt incumbents but redefine entire industries.
Actionable Insights
Identify an expensive, repetitive workflow in an industry and design an AI agent to fully replace it.
Focus on direct sales to CEOs and CFOs, positioning AI as a cost-saving, efficiency-driving investment.
Prioritize hiring engineers skilled in AI automation over building large customer support or sales teams.
Move fast—AI capabilities improve every few months, so iterate rapidly to stay ahead of the competition.
Choose an industry you understand deeply—insider knowledge helps uncover overlooked automation opportunities.
Mind Map
Key Quote
"Every three months things have just kept getting progressively better, and now we're at this point where we're talking about full-on vertical AI agents that are going to replace entire teams and functions in enterprises. That progression is still mind-blowing to me."
Future Trends & Predictions
AI-driven automation is set to reshape the business landscape, with startups replacing entire operational teams rather than just optimizing workflows. As global hiring slows and businesses seek efficiency, lean AI-first companies will dominate, proving that high revenue no longer requires large headcounts. With OpenAI, Anthropic, and emerging players racing to improve foundation models, AI agents will become increasingly sophisticated, accelerating adoption across industries. The rise of AI-powered customer support, sales, and operations means legacy enterprise software giants must adapt or risk being displaced by specialized, automation-first challengers.
Check out the podcast here:
Latest in AI
1. OpenAI has released GPT-4.5, its largest and most advanced language model to date, featuring improved natural conversation abilities, enhanced emotional intelligence, and reduced hallucinations. The model, which supports a 128,000 token context window, will be available to ChatGPT Pro subscribers immediately and to other paid users in the coming weeks, with API access also offered at a higher cost than previous versions.
2. Amazon has unveiled Ocelot, a prototype quantum processor that aims to significantly reduce the resources required for quantum error correction, potentially accelerating the development of practical quantum computers. This first-generation chip, developed by the AWS Center for Quantum Computing, uses a novel architecture focusing on error correction from the start, which could lead to functional quantum computers using as little as one-tenth of the resources needed for conventional quantum error-correction.
3. Meta is planning a major expansion of its AI offerings for business users, aiming to provide AI tools to "hundreds of millions" of businesses, including small enterprises. The company envisions a future where every business, regardless of size, will have an AI agent representing and acting on its behalf, similar to how businesses today have websites and email addresses.
Startup World
1. Tencent has launched Hunyuan Turbo S, its latest AI model, featuring drastically increased speed with response times nearly halved and doubled output per second compared to previous models. The model is built on an innovative hybrid-mamba-transformer fusion architecture, optimizing efficiency and reducing training costs while enhancing reasoning capabilities. Hunyuan Turbo S has shown performance comparable to leading systems like GPT-4o in public benchmark tests, intensifying China's AI race with benchmarks rivaling DeepSeek-V3 and Claude 3.5 Sonnet.
2. Taktile, a SaaS startup based in Berlin, New York, and London, has raised $54 million in a Series B funding round led by Balderton Capital, with participation from existing investors including Tiger Global Management and Y Combinator. The company's no-code software platform enables financial institutions to build and deploy automated decision flows and agents for onboarding, credit, fraud, and compliance, processing hundreds of millions of risk decisions monthly.
3. Swiss startup Unique has secured $30 million in Series A funding led by CommerzVentures and DN Capital, with participation from VI Partners and Pictet Group. The company develops AI agents for banking, asset management, and compliance, aiming to streamline complex financial tasks for institutions in research, due diligence, and KYC operations. Unique plans to use the funding to accelerate its U.S. expansion and further develop its AI platform, which is already used by major financial firms like Pictet, UBP, and Graubündner Kantonalbank.
Analogy
The rise of vertical AI is like the invention of the assembly line—what once required entire teams will soon be automated with precision and scale. Just as SaaS transformed software delivery, AI agents are set to revolutionize workflows, replacing manual processes with intelligent automation. No single AI will dominate; instead, countless specialized agents will streamline industries, much like how factories mechanized production. The companies that spot inefficiencies—like repetitive admin work—will build the next generation of AI-first businesses, turning labor-intensive tasks into seamless, automated operations. The future belongs to those who replace bottlenecks with breakthroughs.
Thanks for reading, have a lovely day!
Jiten-One CerebralAll summaries are based on publicly available content from podcasts. One Cerebral provides complementary insights and encourages readers to support the original creators by engaging directly with their work; by listening, liking, commenting or subscribing.
Reply