In partnership with

Remember when every company needed a website? Then a mobile app? We're at a similar inflection point with vertical AI agents – specialized AI systems designed to revolutionize specific industries and business functions. This isn't just another tech trend; it's a fundamental shift in how software will serve businesses and their customers.

In This Issue

1. The Gold Rush of Our Generation

2. Vertical AI Agents Are Starting To Emerge

3. Looking Ahead

   1. Not subscribed yet?

The Gold Rush of Our Generation

The numbers are mind-blowing. AI agents could contribute up to $15.7 trillion to the global economy by 2030 – more than the current combined output of China and India¹ . This is a fundamental economic restructuring related to scaling the previously scarce resource of knowledge work to industrial scales of efficiency.

Previously, we might ask a knowledge worker for one concept instead we will now be able to get three or even a hundred. The limits on our outputs will be removed. In it’s place the constraints will be on compute and human oversight rather than the actual work deliverables.

AI agent collaboration is in its early stages, offering emerging companies the chance to surpass established players and for startups to become future giants. What role will creators play in designing and developing the next generation of vertical AI agents for applications, operating systems, and corporate intranet infrastructure?

Vertical AI Agents Are Starting To Emerge

• Khanmigo (Education): Khan Academy's AI tutor represents a breakthrough in educational AI. Unlike ChatGPT, Khanmigo doesn't simply provide answers – it guides students through a custom tailored learning process with unlimited patience, leveraging Khan Academy's comprehensive K-12 curriculum² .

• AI Paralegals (Legal): Anthropic and OpenAI are quietly partnering with leading law firms for their data to develop specialized legal assistants. These AI paralegals will go beyond document creation, offering predictive analytics for case outcomes, efficient e-discovery, and automated due diligence for corporate deals. They're designed to augment legal professionals' capabilities while maintaining accuracy and compliance³ .

• Stanford HEA₃RT (Healthcare): Stanford Medical's Healthcare Ethics and AI Research Team (HEA₃RT) takes a unique hands-on approach to agent building by focusing on rigorous evaluation and thoughtful implementation in clinical settings. Their work emphasizes responsible, outcome-driven AI that directly serves healthcare professionals and patients⁴ .

The Big Hurdle In Front of AI Agent Integration

Here's the paradox: compute is getting cheaper, but running AI agents is getting more expensive. Why? Because we're not just paying for processing power – we're investing in intelligence. It is this paradox of cheaper compute mixed with rising costs of more complex AI agent tasks that must be overcome to see these AI agents reach mass adoption.

Today's AI agents are light-years beyond simple chatbots answering FAQs. We're talking about systems that:

• Process multiple data streams simultaneously

• Maintain context across complex conversations

• Collaborate with humans on thorny problems

• Actually learn from institutional knowledge

The price tag reflects this evolution. Like upgrading from interns to senior consultants, these specialized agents need the good stuff to deliver results:

• High-quality document processing

• Image and video analysis capabilities

• Rich contextual understanding

• Deep integration with human expertise

Making These Vertical Agents Work At Scale

The game is all about smart system design. We're building architectures that:

• Balance raw processing power with intelligent resource allocation

• Batch similar requests to maximize efficiency

• Use advanced caching and vector search for speed

• Optimize context windows for maximum value per token

• Seek human feedback at key moments in these workflows to keep impact high and compute prices low

Creating Real Expertise

Success isn't about minimizing costs – it's about maximizing impact:

• Target high-value problems where AI expertise makes a real difference

• Design workflows that combine human insight with AI processing power

• Create strategic touch-points for human-AI collaboration

• Build measurement systems that track both immediate savings and long-term impact

When you nail this, every interaction between humans and AI counts. No wasted tokens, no unnecessary round-trips – just pure value delivery.

Building Vertical AI Agent Expertise

What can we do as creative technologists working with AI agents to learn to build these systems? We can share the themes and approach Anastasia and I are taking:

1. Master the Fundamentals

• Deep dive into what LLMs can (and can't) do

• Get cozy with vector databases and embeddings

• Level up your prompt engineering game

2. Pick Your Vertical

• Choose an industry you're genuinely excited about

• Map out where AI can make the biggest impact

• Connect with the experts who know the domain inside out

3. Build Like Crazy

• Start small but think big creating agents to help with your tasks while considering how other professionals like you might want to interact with a team of agents helping with their jobs too.

• Experiment with different tools that are being released to understand the pros and cons

• Share what you learn – the community needs it to uncover the best practices and dead-ends. This is why we write this newsletter.

4. Connect with Domain Experts

• Connect with knowledgeable experts. Most experts are willing to share a quick half-hour to discuss a specific professional challenge and how they might use a new AI agent.

• Learn the unwritten rules and hidden workflows of domain experts to build it into your new AI agents.

• Build relationships that inform your AI development. Circle-back with experts sharing your new AI agents to keep them in the loop with what you’ve built for testing and continuous feedback.

Looking Ahead

The vertical AI revolution focuses on solving real human problems in specific industries by combining technical skills with deep domain knowledge to create next-generation AI agents.

The tools are here. The market is ready. The potential is massive. The question isn't whether vertical AI will transform software – it's who's going to lead the charge.

What We’re Reading & Experimenting With This Week

Stay up-to-date with AI

The Rundown is the most trusted AI newsletter in the world, with 1,000,000+ readers and exclusive interviews with AI leaders like Mark Zuckerberg, Demis Hassibis, Mustafa Suleyman, and more.

Their expert research team spends all day learning what’s new in AI and talking with industry experts, then distills the most important developments into one free email every morning.

Plus, complete the quiz after signing up and they’ll recommend the best AI tools, guides, and courses – tailored to your needs.

Sign up to start learning.

Suggestion box

Forward it to a friend and have them signup here.

Until next time, keep innovating and stay curious!

Jason and Anastasia 

Not subscribed yet?

This newsletter is part of our ongoing series exploring the future of specialized AI through a human-centered lens. Subscribe to join a community of designers and developers shaping purposeful AI agents.

1  Evidence of Need for Skilling In The Areas of Artificial Intelligence and Machine Learning

2  Khanmigo: On-demand AI-powered support for education

3  The Future of Legal Assistants: AI for Enhanced Productivity

4  Stanford HEA₃RT: Healthcare Ethics and AI Research Team