This Revolutionary AI Business Model Will Make Millionaires in 2025

Key Concepts

• Agents as a Service (AAS): A new AI business model where AI agents automate entire processes, unlike SaaS which only provides tools.
• Vertical Agents: AI agents designed for a specific role or function, pre-trained on general processes.
• Horizontal Agents: AI agents that are not restricted by specific use cases and can be customized for any niche, role, or function.
• Standard Operating Procedures (SOPs): Documented processes that guide how tasks are performed within an organization.
• Minimum Viable Product (MVP): A version of a product with just enough features to satisfy early customers and provide feedback for future development.
• Evals: A system for tracking and evaluating the performance of AI agents to enable self-improvement.

1. The Rise of Agents as a Service (AAS)

• AAS as a Potential Replacement for SaaS: The video posits that AAS could surpass or replace Software as a Service (SaaS), a $300+ billion market.
• Shift from On-Premise to SaaS: The transition from software sold on CDs to the SaaS model was enabled by XML HTTP request (Ajax).
• Enhanced Interactivity with LLMs: Large Language Models (LLMs) allow for more natural interaction with software, potentially eliminating the need for browsers.
• AAS Automates Entire Processes: Unlike SaaS, which provides tools, AAS automates entire processes, making it more scalable. Example: A marketing agent handles lead generation, appointment scheduling, follow-ups, and CRM updates, eliminating the need for a marketing person.
• Easier to Build Than Traditional Software: AI agents can directly interact with databases, reducing the need for complex APIs and user interfaces.
• Market Growth: In 2024, enterprises spent $1.2 billion on vertical agents, a 12x increase from the previous year.

2. Vertical vs. Horizontal AI Agents

• Vertical Agents:
  • Definition: Designed for a specific role or function. Example: Harvey AI for law firms.
  • Pre-training: Can be pre-trained on general processes, allowing for faster setup.
  • Customization: Harder to customize for specific business processes due to pre-training.
  • ROI: Offer a steep initial ROI but performance may plateau.
  • Ease of Pitch: Easier to pitch and build due to targeting a specific customer.
• Horizontal Agents:
  • Definition: Not restricted by specific use cases, adaptable for any niche. Example: Agency swarm.
  • Training: Require training from scratch.
  • Customization: Easily customized for any process.
  • ROI: Require significant initial investment but can achieve higher ROI due to greater customization.
• Business Preference: Most business owners prefer vertical solutions due to the quicker path to 80% performance with 20% effort.

3. Real-World Examples of Vertical Agents

• 11x: An AI SDR and sales rep agent automating go-to-market strategies, replacing platforms like Salesforce and Apollo. Raised $50 million in Series B funding, valuing the company at $350 million.
• Carmen: An agent for construction project managers automating administrative tasks.
• Normi: An agent for regulatory compliance teams evaluating content and actions against regulations.
• Devon: A $500/month development agent that works in Slack.

4. Building a Vertical AI Agent: Key Components

• Data: High-quality, internal data is crucial for training, evaluation, and fine-tuning.
• Industry-Specific Expertise: Deep understanding of the target customer and their SOPs is essential.
• Resources: Funding is not mandatory; time and effort can suffice.

5. Methods for Building Vertical Agents

• Framework: Using agentic frameworks like agency swarm, crew AI, link chain, and autogen.
  • Pros: Saves time and effort during development.
  • Cons: Requires some development experience.
• Horizontal Platform: Building on platforms like Google Cloud vertic AI agent Builder or a s Bedrock agents.
  • Pros: Scalability without managing servers.
  • Cons: Can be expensive at scale.
• Custom Coded Solution: Building everything from scratch.
  • Pros: Complete control.
  • Cons: Requires significant technical expertise and development effort.

6. Pricing Models for Vertical Agents

• Licensing: Clients pay a one-time setup fee or monthly subscription.
  • Pros: Easy to pitch.
  • Cons: Doesn't account for value generated or usage.
• Usage-Based: Charging based on token or message consumption.
  • Pros: Scalable.
  • Cons: Requires constant usage.
• Outcome-Based: Charging per result (e.g., appointment booked, lead generated).
  • Pros: Clients can easily evaluate pricing.
  • Cons: N/A
• Hybrid: Combining multiple pricing strategies.
  • Pros: Benefits of different models.
  • Cons: Harder to estimate long-term costs.

7. Roadmap for Leveraging AAS in 2025

1. Find Your Niche: Identify industries you know well and have unique insights into.
2. Identify a Suitable Problem: Find a recurring problem that companies struggle to automate and that traditional automation tools can't solve.
3. Sell First: Find a client before building the agent to reduce upfront risk.
4. Build an MVP: Create a minimum viable product tailored for the initial client, collecting necessary data and feedback.
5. Productize: Identify consistent features and make everything else easily modifiable from a config file or template.
6. Evaluate: Set up evals to track the agent's performance and enable self-improvement.
7. Scale: Increase marketing spend, hire more people, and deliver as many agents as possible.

8. Interview with Chase from Infinite AI

• Solution: Chase built a vertical AI voice solution that integrates with CRMs and automations to handle everything from lead generation to customer follow-up.
• Key Features: AI agents can call and text leads, update CRMs, book appointments, and move leads through pipelines without human intervention.
• Origin of Idea: Chase's experience in the solar panel industry revealed the need for scalable automation in lead management.
• Technical Expertise: Chase had some self-taught experience with CRMs and automations but no formal IT or software engineering background.
• Templatization: The solution is largely templatized, allowing for quick customization for different clients.
• Pricing Model: Offers both "build and release" and "growth partner" models, with the latter involving a revenue share.
• Revenue: Generates an average of $30,000 to $60,000 per month in recurring revenue, with a best month close to $100,000.
• Demo: A live demo showcased the capabilities of Grace, an AI sales assistant, including multilingual support, rap performance, and seamless transfer to a sales agent.

9. Future of Vertical vs. Horizontal Agents

• Coexistence: Both vertical and horizontal agents will have their place in the future.
• Horizontal Platforms Building Vertical Agents: Horizontal platforms will eventually be able to build vertical agents for any use case from a single prompt.
• Self-Improvement: Agents will be able to set up their own evals and self-improve.
• Business Models: Some will prefer vertical, some horizontal, and many will combine both to start businesses without any employees.

10. Synthesis/Conclusion

The video presents a compelling case for Agents as a Service (AAS) as the next major business model, potentially surpassing SaaS. It highlights the differences between vertical and horizontal AI agents, providing real-world examples and a detailed roadmap for building and scaling a vertical AI agent business. The interview with Chase from Infinite AI offers practical insights and demonstrates the potential for generating significant revenue with a well-executed vertical AI solution. The video concludes by suggesting that both vertical and horizontal agents will coexist, with horizontal platforms eventually enabling the creation of vertical agents from simple prompts.