- One more Thing in AI
- Posts
- Edition #33: Are AI Agents Going Mainstream in 2025?
Edition #33: Are AI Agents Going Mainstream in 2025?
Latest edition of One More Thing in AI Newsletter.
Renjit Philip
December 12, 2024 • Estimated Reading Time: 13 minutes
AI may be faster than us. AI may be smarter than us, but it can’t be more human than us... AI agents can be wonderfully helpful."
Date: 12-Dec-2024
Hey AI enthusiast,
Welcome to the latest edition of the One More Thing in AI newsletter!
Today’s edition is all about AI Agents. Agents are set to transform productivity and work as we know it. Imagine intelligent autonomous agents that deliver high-quality work at your command! That is the promise of AI Agents.
Commentators from YCombinator have stated that vertical agents could have 10x the TAM of SaaS companies. Let us find out why.
Best,
Renjit
In this edition:
What is an AI Agent?
Think of it like this: instead of just giving you a direct answer like a search engine or a simple chatbot, an AI agent is like a personal assistant who can use different tools and resources to get things done.
Here's what makes AI agents special:
Reasoning: Unlike regular computer programs that follow pre-programmed rules, AI agents can actually think and reason like a human. They can break down a complex task into smaller, more manageable steps and determine the best way to achieve the desired goal.
Action: AI agents aren't just thinkers; they're doers! They can interact with programs, apps, and websites to gather information, process data, and control other systems. It's like having your robot that can search the web, send emails, book appointments, and even write code for you.
Memory: AI agents can remember past interactions and use that information to make better decisions in the future. They're constantly learning and improving their skills based on their experiences.
Tool Use: AI agents have access to a whole toolbox of external programs and APIs that they can use to accomplish their goals. These tools include search engines, databases, calculators, code execution engines, and AI models. The agent decides which tools to use and how to use them based on the specific task.
Do you want to reach a high-quality audience?
To sponsor this Newsletter, subscribed to by startup founders and business leaders:
Market Map for the AI Agent space
The map tells you about the infrastructure code and framework needed to make AI Agents come to life. I am optimistic about the prospects of picks and shovel type of businesses and the hosting services, agent frameworks, storage sectors etc., fit the bill.
Market Map for AI Agents (source Adam Silverman on X.com)
AI Agent Startups that Caught my eye
Maven AGI (Customer Support): Offers generative AI agents for customer support, creating personalized responses. Maven AGI's platform adapts to customer needs and dynamically adjusts responses. It has received $28M in funding.
Cognition.ai (Software Development): Pioneers in agent-based platforms for software development, autonomously building websites and videos from user prompts.
FACTORY (Software Development): Created an AI-driven lifecycle automation tool with AI agents for software deployment and testing. It automates tasks and integrates with CI/CD pipelines for continuous delivery.
11x.ai (Sales): Leverages data insights to predict sales patterns, offering recommendations for sales teams from lead qualification to nurturing (constantly in autopilot mode). This allows teams to focus on high-impact interactions.
Clay (Sales): Integrates with CRM systems to provide sales teams with enriched data insights on leads and clients. It has AI agents that automatically research prospects. Clay recently reached a $500M valuation.
Observe.ai (Customer Support): Focuses on AI integration in call centers, offering real-time transcription and sentiment analysis tools to guide agents during live calls. Their platform helps optimize customer service and improve agent performance.
Decagon (Customer Support and Supply Chain Management): Builds AI agents to transform support workflows and optimize logistics. Decagon has raised $100M.
MultiOn (Browser Agent): A web agent that uses vision transformers trained on software interfaces and codebases. This allows the agent to "understand" web components, automate web browsing, and perform visual UI actions and text entry.
Sema4 (Enterprise Agent): Their invoice reconciliation agent showcases Sema4's core processing capability. The same core technology can be applied to various data validation tasks across finance, procurement, and operations.
Make your voice heard!
I would love to know more about what you like and don’t like, specifically in this Newsletter. Please take this 2-minute survey to help me produce more useful content for you.
Agent FAQs
1. What are AI Agents, and how are they different from traditional bots?
AI agents are sophisticated software programs that leverage artificial intelligence, particularly large language models (LLMs), to achieve complex goals requiring multiple steps. Unlike traditional bots that operate on pre-programmed rules, AI agents possess reasoning abilities, enabling them to adapt to changing circumstances, make decisions, and interact with various tools and data sources to fulfill their objectives.
2. What are the key characteristics that define an AI agent?
AI agents are characterized by:
Reasoning: They can break down complex goals into smaller steps and devise execution plans.
Action: They interact with external tools and APIs to gather information, manipulate data, and perform tasks.
Memory: They store and retrieve information from past interactions to inform future actions.
Learning: They can learn from experience to improve their performance over time.
3. How do AI agents utilize "tools" to accomplish tasks?
AI agents can access various "tools," essentially external programs or APIs extending their capabilities. These tools can include:
Search engines for retrieving information.
Databases for storing and accessing data.
Calculators for performing mathematical operations.
Code execution engines for running code.
Other specialized AI models for tasks like translation or image recognition.
The agent decides which tools to use and how to use them based on the task.
4. What are the different types of AI agents, and how are they categorized?
AI agents can be categorized based on their:
Domain Specificity
Horizontal Agents: Possess broad capabilities applicable across multiple domains.
Vertical Agents: Specialized for specific industries or departments. The bull case for vertical AI agents could surpass SaaS because, unlike SaaS, which requires an operations team to manage workflows, AI agents can replace SaaS software and significantly reduce payroll costs.
Since payroll remains one of the largest expenses for companies compared to the relatively small cost of software, AI agents have the potential to revolutionize both areas, creating a much larger impact.
This is why commentators at YC have said that vertical AI Agents could have a 10x TAM compared to current SaaS companies.
LLM Autonomy:
Highly Autonomous Agents Can independently plan and execute complex workflows.
Guided Agents: Require more human input and oversight.
5. What are some challenges and considerations for implementing AI agents in the enterprise?
Data Security and Privacy: Safeguarding sensitive data handled by agents is crucial.
Integration with Existing Systems: Seamless integration with current enterprise software is essential.
Explainability and Transparency: Understanding the reasoning behind an agent's actions is essential for trust and accountability.
Human Oversight and Control: Maintaining human oversight to ensure ethical and responsible AI agent behavior.
Forward to a friend
Ready to build an AI Agent in under 5 minutes?
Build AI Projects Without the Hassle Using AgentStack
Starting AI projects doesn’t have to be complicated—AgentStack simplifies the setup process.
Interactive Tutorials Available: Follow the step-by-step guide to build a web scraper agent, which is perfect for learning the basics. See the link here>>
Another helpful framework for building AI Agents is Crew.ai. Give that a go as well.
Imagine orchestrating a team where each member excels in their role, working together seamlessly to achieve complex goals.
CrewAI brings this vision to life by enabling the creation of specialized AI agent teams that collaborate effectively.
AI Agent Startup Ideas for you
All right, the whole AI agents space has fired up my imagination. Here are a few startup ideas I am toying with. Let me know what you think.
AI-Powered Contract Review and Negotiation Agent (Legal Tech): This agent would automate the tedious and time-consuming contract review and negotiation process.
The legal tech market presents a substantial TAM. For instance, DocuSign estimated its TAM to be over USD 50 billion in 2021, while LegalZoom identified a Serviceable Addressable Market (SAM) of USD 48.7 billion.
Distribution could be achieved through partnerships with law firms or corporate legal departments. Initial funding might be higher due to the need for specialized legal expertise and training data. Still, the potential for cost savings and efficiency gains in the legal profession is significant.
AI-Driven Code Review and Bug Detection Agent (DevOps): This agent would help developers identify and fix bugs more efficiently, leading to faster release cycles and improved software quality.
Distribution could leverage existing developer communities and platforms like GitHub.
The AI Code Tools Market valuation is predicted to surpass USD 30.1 billion by 2032, as reported in a research analysis by Global Market Insights.
AI-Powered Personal Shopping Assistant: Imagine an AI agent acting as your personal shopper in the digital world. This agent would:
Learn your style, preferences, and budget by analyzing your past purchases, browsing history, and social media activity.
Proactively suggest products and deals that match your taste, saving you time and effort from endless scrolling through online stores.
Negotiate the best prices on your behalf by monitoring price fluctuations and taking advantage of flash sales or discounts.
Help manage your shopping budget and prevent impulse purchases by providing personalized spending insights.
AI-Powered Hyper-Personalized Trip Planner: This AI agent would take travel planning to the next level by creating bespoke itineraries that cater to your unique interests and needs. It would:
Learn your travel preferences (adventure vs. relaxation, budget, the preferred mode of transportation, dietary restrictions, etc.) through a conversational interface.
Scour the web for flights, accommodations, activities, and local experiences, considering real-time pricing and availability.
Generate detailed, visually appealing itineraries that include suggested flights, hotels, restaurants, must-see attractions, hidden gems, and even personalized recommendations based on weather forecasts and local events happening during your trip.
Continuously monitor for changes (flight delays, price drops, event cancellations) and provide proactive updates and alternative suggestions, ensuring a seamless travel experience. You can read this article on McKinsey.com for more insights into this sector.
AI-Powered Financial Planning and Investment Advice Agent (FinTech): This agent would help individuals create personalized financial plans and make informed investment decisions on autopilot.
This agent would have access to your trading account, bank accounts, crypto wallets, and so on.
Distribution could be achieved through partnerships with financial institutions or directly to consumers. Funding needs could be moderate to high, as regulatory compliance and financial expertise would be crucial.
According to Grand View Research, the market size for AI applications in fintech alone is expected to increase from US$ 9.45bn in 2021 to US$ 41.16bn by 2030 at a compound annual growth rate of 16.5 percent.
Please let me know if you plan to build any of the above, and I will be happy to work together. You can book time with me via my booking link on LinkedIn or by replying to this email.
Reply