Death Comes to the Middleman
Disintermediation and the Future of AI Agents | Claude Learns to Use a Computer | OpenAI's Swarm Framework | Willy Loman Meets a Different Ending?
👋 Hey, I'm Bryan. Welcome to Code & Context, where we explore how to survive and thrive in the age of artificial intelligence. Through essays, commentary, and AI code experiments, I share insights on AI, culture, and creativity – helping you navigate the future that's already here.
“Because with culture, we can beat the machines.”
Subscribe now to join our community of forward-thinkers shaping our AI-infused future. Get thought-provoking essays, practical AI insights, and early access to tools from Intertwine AI.
Disintermediation is a beautiful word. It rolls off the tongue with the promise of freedom, of cutting through red tape, of finally getting things done. At its core, it means eliminating the middleman—those gatekeepers who create layers between what we want and what we can get, who extract value while often making processes more complex rather than simpler. And with the rise of AI agents—intelligent software that can negotiate, analyze, and advocate on our behalf—this idea is finally becoming reality.
I've been thinking a lot about this lately, ever since my partner's neighbor's house caught fire over the summer. Smoke, that insidious byproduct of disaster, knows no property lines. It seeped into every nook and cranny of her home, leaving a layer of soot and an acrid smell that seemed to mock the very idea of safety and security. You'd think that in moments like these, insurance—that promise of protection we dutifully pay for month after month—would swoop in like a superhero, cape fluttering in the wind.
Instead, Liberty Mutual (yes, I’m calling out your crap, Limu Emu) transformed from a beacon of hope into a labyrinth of frustration. Each interaction felt designed to confuse and delay. Simple questions about coverage met with contradictory answers. Estimates for smoke damage remediation arrived weeks late, each lowball number seemingly pulled from thin air. As months passed with her home still in disarray, I couldn't help but wonder: In an age of AI agents that can understand complex documents, make phone calls, and negotiate on our behalf, why does this process feel like it's stuck in the last century?
The Ghost of Willy Loman
All this bureaucratic gatekeeping—the dance of forms and phone calls and perpetual delays—reminds me of Arthur Miller’s Death of a Salesman. The play’s protagonist, Willy Loman, embodies both the tragedy and the necessity of middlemen in a world before artificial intelligence. A traveling salesman desperately clinging to an outdated version of the American Dream, Willy survives by inserting himself between manufacturers and customers, adding a layer of human charm to every transaction. But as markets change and his personal connections fade, he finds himself increasingly irrelevant, unable to compete with more efficient systems.
What makes Willy’s story particularly haunting is how it foreshadows our current moment of technological disruption. Like insurance adjusters and other modern middlemen, Willy’s power comes from information asymmetry—his knowledge of products, prices, and people that his customers can’t easily access on their own. He transforms this asymmetry into value through a combination of charm, persistence, and sometimes deception. When the asymmetry fails him, when customers become too savvy or competition too fierce, his world crumbles.
Today’s bureaucratic middlemen—the insurance adjusters, administrators, and gatekeepers of all stripes—are modern-day Willy Lomans. They thrive on complexity and controlled access to information. They maintain their positions not by adding genuine value but by making themselves seemingly indispensable to navigating complex systems.
The rise of AI agents presents two possible futures for such middlemen. In one, they embrace the technology, using AI to augment their capabilities rather than replace them. Imagine a Willy Loman armed with AI tools that could analyze market trends, optimize routes, and provide real-time insights. He might transform from a desperate huckster into a genuine consultant, using technology to deliver real value to his customers.
But there’s another, more likely future: one where AI agents simply eliminate the need for middlemen altogether. When every customer has their own AI advocate—one that can analyze products, negotiate prices, and cut through bureaucratic complexity—what role is left for the Willy Lomans of the world?
This is the real meaning behind our title, “Death Comes to the Middleman.” The tragedy of Willy Loman wasn’t just that he failed to adapt—it was that his entire way of doing business, his very role in the economic ecosystem, was destined for extinction. Today’s middlemen face the same existential threat, but at an unprecedented scale and speed, driven by artificial intelligence that can replicate and exceed their core functions.
Seeing is Believing
If you’re skeptical about AI agents being able to handle real-world interactions, OpenAI’s recent developer conference might change your mind. In a series of demonstrations that felt like science fiction come to life, they revealed just how close we are to a world without middlemen.
Most striking was their demonstration of an AI agent making an actual phone call to order strawberries. The AI didn’t just follow a script—it engaged in a natural conversation, asking about prices, negotiating delivery times, and even discussing chocolate coating options. The interaction was so smooth that it was hard to tell one participant wasn’t human. Now imagine that same technology applied to calling insurance adjusters, contractors, or claims departments. Instead of spending hours on hold or playing phone tag, your AI agent could handle these conversations for you, documenting everything and ensuring nothing falls through the cracks.
In another demo that hints at AI’s growing capabilities, they showed an AI agent controlling a drone in real time, writing all the necessary code on the fly. While flying drones might seem far removed from insurance claims, the underlying technology—an AI that can understand complex systems and take precise actions in real time—could revolutionize how we deal with bureaucracy. Picture an AI that could navigate through online claim systems, fill out forms, and even catch errors or inconsistencies as they happen.
These weren’t just carefully scripted demos. Weeks before, OpenAI introduced their latest AI models, codenamed “o1,” which they describe as having PhD-level problem-solving abilities. It uses a “chain of thought” reasoning technique to break down complex problems into manageable pieces—exactly the kind of thinking needed to untangle something as complicated as an insurance claim.
AI “Agents” You Can Use Today
While we’re still waiting for AI agents that can order strawberries on our behalf, several tools available right now could have helped our friend navigate her insurance nightmare:
CrewAI: Orchestrating AI Agents for Complex Tasks
CrewAI is a platform that enables the creation and deployment of multiple AI agents working collaboratively to automate complex workflows. In the context of insurance claims, CrewAI might help:
• Automate Claims Processing: AI agents can handle tasks such as claim intake, document verification, and status updates, reducing manual effort and accelerating processing times.
• Data Extraction and Aggregation: The system efficiently extracts and consolidates data from various sources, minimizing errors and ensuring consistency.
• Customer Support Automation: AI agents manage customer inquiries, providing timely responses and resolving common issues, thereby enhancing customer satisfaction.
What makes it revolutionary: CrewAI’s platform reduces operational costs, minimizes human errors, and improves customer satisfaction by automating complex insurance processes.
ChatGPT Plus with Code Interpreter
While not specifically designed for insurance claims, this tool could help:
• Break down insurance policy language into plain English
• Compare repair estimates for inconsistencies
• Draft professional correspondence with insurance adjusters
• Analyze smoke damage documentation and costs
• Generate detailed timelines of events and communications
What makes it revolutionary: It turns complex insurance documents and data into understandable, actionable information.
This AI research assistant could:
• Find relevant insurance regulations and consumer protection laws
• Research similar cases and their outcomes
• Gather data about standard smoke damage remediation costs
• Identify potential leverage points in negotiations
What makes it revolutionary: It helps close the information gap between individual consumers and large insurance companies.
These tools work best when used together. For example, you might:
Use ChatGPT to analyze your insurance policy and identify key coverage points.
Use Perplexity to research similar cases and standard costs.
Use CrewAI to automate claims processing and manage customer interactions efficiently.
The Future is Closer Than You Think
While these current tools are already impressive, the work being done at frontier AI labs like OpenAI and Anthropic offers an even more tantalizing glimpse of our AI-empowered future.
OpenAI’s experimental “Swarm” framework allows multiple AI agents to work together, handing off tasks between them like a well-coordinated team. Imagine not just one AI advocate on your side, but a whole squad of specialized agents—one analyzing your insurance policy, another negotiating with adjusters, and a third coordinating with contractors.
Meanwhile, Anthropic has made breakthroughs that feel almost magical. Their latest Claude model can actually control computers directly—moving cursors, clicking buttons, and typing information. More importantly for consumers dealing with bureaucracy, it can analyze complex PDFs and spreadsheets with unprecedented sophistication. Imagine an AI assistant that could not only read your insurance policy but also compare it line by line with standard policies, analyze repair estimates in detail, and flag any discrepancies or unusual terms.
The implications are profound. When these capabilities mature and become widely available, the traditional advantages of large corporations—their armies of adjusters, complex documentation, and bureaucratic processes—could become obsolete overnight. Every consumer could have access to AI agents that understand complex documents as well as any expert, can operate computer systems to navigate bureaucratic processes, and can think through problems with the thoroughness of a trained professional.
We’re not quite at the level of AI ordering strawberries for everyone yet, but these developments from the frontier labs show us that the future we imagined—where AI agents level the playing field between consumers and corporations—isn’t just possible. It’s inevitable.
The Revolution at Hand
The beauty of this AI revolution lies not just in its potential for efficiency but in its power to democratize expertise. When everyone has access to AI agents that can analyze contracts, negotiate terms, and manage complex interactions, the traditional advantages of large corporations—their armies of adjusters, agents, and administrators—begin to crumble.
Imagine a world where every consumer has their own AI advocate, leveling the playing field in dealings with insurance companies, banks, healthcare providers, and other institutions that have historically held the upper hand through information asymmetry and bureaucratic complexity.
This isn’t just about making processes more efficient—it’s about fundamentally rebalancing power. It’s about creating a future where the Willy Lomans of the world don’t need to rely on charm or deception, and where consumers don’t need to feel powerless in the face of corporate bureaucracy.
The death of the middleman may be upon us, but from its ashes, a new paradigm of consumer empowerment is poised to rise. It’s a future that promises to be more efficient, more transparent, and perhaps, paradoxically, more human. Because when we strip away the bureaucracy, the information asymmetry, and the need for charm and personal connections, what we’re left with is the essence of what matters: fair treatment, clear communication, and problems solved.
The revolution isn’t coming. It’s here. And it’s open to everyone who’s ready to embrace it.