How has the evolution of technology influenced the rise of SaaS and now AI agents?

Key technological advancements like XML HTTP request enabled rich internet applications, fueling the SaaS boom. Similarly, LLMs represent a new computing paradigm enabling fundamentally different applications like vertical AI agents.

Why didn't large tech companies like Google create the disruptive products like Uber or Airbnb?

Incumbents often avoid risky ventures in new markets due to the 'Innovator's Dilemma' – the fear of endangering their existing profitable revenue streams.

What makes vertical SaaS products successful compared to broad enterprise suites like Oracle?

Vertical SaaS products offer a 10x better user experience by deeply understanding specific domain nuances, unlike broad suites that are often 'jack of all trades, master of none'.

How is the hiring strategy for startups changing with the advent of LLMs?

Startups are shifting from hiring traditional roles like customer success managers to hiring more software engineers who can leverage LLMs to automate bottlenecks and reduce the need for large teams.

What is the best way for founders to identify a lucrative vertical for an AI agent startup?

Look for boring, repetitive administrative tasks or 'butter-passing jobs' that are ripe for automation. Direct experience or a personal connection to the domain can also be a strong indicator.

How can AI agents like MTIC overcome the friction of replacing existing teams?

By directly targeting the function to be replaced (e.g., QA testing) and selling to those who benefit from the reduction (e.g., engineering heads), rather than trying to incrementally improve the roles they aim to supplant.

What is the significance of "rocks can read" in the context of AI and management?

This refers to LLMs' ability to process and understand vast amounts of information, empowering managers to extend their reach and context window, potentially increasing the scale of firms they can oversee.

Key Moments

Vertical AI Agents Could Be 10X Bigger Than SaaS

Y Combinator

Science & Technology5 min read43 min video

Nov 22, 2024|964,894 views|17,375|579

YC Y Combinator Vertical AI Agents OpenAI Anthropic Claude GPT-4 ChatGPT

Save to Pod

Key Moments

TL;DR

Vertical AI agents, disrupting SaaS by automating teams, could create $300B+ companies.

Key Insights

Vertical AI agents represent a significant evolution beyond general foundational models, capable of replacing entire teams and business functions.

The emergence of multiple strong AI foundation models (like OpenAI and Claude) fosters competition, creating a richer ecosystem for startups and consumers.

The history of SaaS, catalyzed by technologies like AJAX, shows a parallel to the current AI revolution, moving software from desktop to accessible web/app-based solutions.

Incumbents often fail to capture value in new technological paradigms due to focusing on existing revenue streams (innovator's dilemma), creating opportunities for startups.

Vertical AI agents can offer 10x better user experiences and potentially displace a larger market than SaaS by automating not just software but also significant portions of human labor (e.g., payroll).

Identifying 'boring, repetitive admin tasks' or 'butter-passing' jobs is a strong indicator for potential high-value vertical AI agent startups.

THE ASCENSION OF VERTICAL AI AGENTS

The rapid progression of AI models has led to the development of vertical AI agents, poised to replace entire teams and functions within enterprises. This evolution from foundational models to specialized agents signifies a paradigm shift. The increasing competition among AI model providers, such as OpenAI and Claude, is crucial for fostering a healthy market, offering choices to consumers and opportunities for founders. This competitive landscape is accelerating innovation and market growth.

LESSONS FROM THE SAS BOOM

The current AI revolution draws parallels with the Software as a Service (SaaS) boom. The key technological catalyst for SaaS was AJAX (Asynchronous JavaScript and XML), enabling rich internet applications that mimicked desktop experiences. This shift from installable software to web-based services paved the way for companies like Google Maps and Gmail. Paul Graham's early work with web applications also foreshadowed this transition, demonstrating the potential of browser-based software.

INCUMBENTS' DILEMMA AND STARTUP OPPORTUNITIES

Historically, incumbents focused on existing profitable products often miss opportunities in emerging tech paradigms. This 'innovator's dilemma' explains why established tech giants didn't dominate categories like ride-sharing (Uber) or accommodation (Airbnb). These ventures often carried regulatory risks that large companies were unwilling to undertake. Similarly, in the B2B SaaS space, incumbents didn't pursue the highly specialized needs of every vertical, leaving room for numerous focused SaaS companies to emerge.

THE 10X POTENTIAL OF VERTICAL AI AGENTS OVER SAS

Vertical AI agents are predicted to be significantly larger than SaaS because they can automate not only software workflows but also a substantial portion of human labor. While SaaS replaced some manual processes, AI agents can handle complex tasks, data entry, approvals, and even entire job functions. This means they could disrupt not only software spending but also a larger chunk of payroll expenses, leading to highly efficient companies with smaller human teams.

THE DISRUPTION OF ENTERPRISE AND B2B SOFTWARE

The traditional enterprise software market, characterized by complex installations and often poor user experiences (e.g., Oracle, SAP), is ripe for disruption. Vertical AI agents can offer superior, specialized solutions that provide a 10x better user experience. This contrasts with early web applications that were initially perceived as less sophisticated. The ability of AI agents to deeply understand and automate specific industry tasks, such as customer support, QA testing, or medical billing, sets them apart from general-purpose tools.

IDENTIFYING OPPORTUNITIES IN BORING TASKS

A common thread among successful vertical AI agent startups is their focus on automating 'boring, repetitive admin tasks' or 'butter-passing' jobs. These are often areas that humans find tedious and that have high churn rates in operational roles. Examples include automating customer collection, QA testing, or even the process of bidding on government contracts. Founders who identify these specific, repetitive problems within a niche are well-positioned to build billion-dollar companies.

THE EVOLVING ROLE OF ENGINEERS AND MANAGERS

The rise of AI impacts how companies are structured and managed. Instead of scaling headcount linearly with revenue, AI agents can automate tasks, allowing for growth with fewer employees. This suggests a shift towards hiring highly skilled software engineers who can build and leverage AI to create efficiencies. Furthermore, AI tools can enhance managerial capabilities, extending their scope and ability to manage larger or more complex organizations, potentially altering the optimal size of firms.

SPECIALIZATION AND THE FUTURE OF ENTERPRISE SOLUTIONS

The market is moving towards hyper-specialization rather than a single, all-encompassing AI platform. While general-purpose AI assistants are an obvious opportunity, incumbents are likely to compete fiercely there. The real value lies in vertical AI agents tailored to specific industries or functions. Enterprises, already trained on the benefits of specialized SaaS solutions, are receptive to startups offering powerful, niche AI agents that deliver significant improvements over existing broad platforms.

VOICE AND THE ACCELERATING PACE OF AI INNOVATION

Voice technology is a rapidly advancing area within AI agents, demonstrating dramatic improvements in realism and latency. Applications like AI-powered calling for debt collection or customer support are moving from basic demos to scalable solutions that can replace human agents. The ease of building on AI platforms (like voice infra companies) allows for rapid deployment, though retaining customers as underlying APIs evolve remains a challenge. This rapid innovation highlights the accelerated progress seen in AI over the past few years.

THE DIRECT SALE: SELLING TO TEAMS NOT BEING REPLACED

When selling vertical AI agents, particularly in enterprise settings, it's crucial to target teams or individuals who are not directly threatened by the technology. Selling solutions that replace entire departments can lead to sabotage from within. Companies are increasingly focused on selling to higher levels of management or to engineering teams who see the efficiency gains, rather than to the end-users whose jobs might be at risk. This strategy avoids internal friction and facilitates adoption.

Mentioned in This Episode

●Software & Apps

●Companies

●Concepts

●People Referenced

Vertical AI Agents: Key Considerations

Practical takeaways from this episode

Do This

Identify boring, repetitive administrative tasks as prime targets for AI agent startups.

Focus on specialized, vertical solutions rather than general-purpose AI.

Consider founders with direct experience in the domain they are targeting.

Leverage AI to empower managers and extend their capabilities.

Look for opportunities where AI can completely replace a function, not just make it more efficient.

Avoid This

Don't overlook the potential for incumbents to win in broad, general-purpose consumer AI applications.

Avoid building AI agents that directly threaten the jobs of the team you need to sell to.

Don't underestimate the complexity required beyond simple LLM prompting for enterprise-grade solutions.

Don't assume that early SaaS successes translate directly to early LLM successes in all categories.

Common Questions

The argument is that vertical AI agents not only replace the SAS software itself but also the operational teams and a significant portion of payroll needed to run that software, leading to potentially 10x larger companies.

Topics

Vertical AI Agents Disruptive Technology B2B Startups

Mentioned in this video

Companies

Qualtrics

A company in the survey space that is likely to be disrupted by specialized LLM agents like Outset.

Viaweb

Cited as an early online store and potentially the first SaaS app, though limited by technology at the time.

PowerOps

An AI customer support agent company that exemplifies the need for complex software beyond simple prompting to replace human support teams.

SweetSpot

A company building an AI agent to bid on government contracts, inspired by a friend's tedious manual process.

Concepts

Coase's Theory of the Firm

The economic theory suggesting firms grow to a point of inefficiency and specialize, which is relevant to understanding firm limits and the potential impact of AI on managerial capabilities.

Dunbar's Number

The cognitive limit to the number of people with whom an individual can maintain stable social relationships, which AI may help extend by processing more information.

Software & Apps

Claude

NetSuite