A One-man media company (Arming Myself with AI Agents and How I did it)

I can write a newsletter draft and tweets from a single sentence.

How is this possible?

That’s what we’ll be talking about today. I’m going to breakdown how I made an AI Agent pipeline that does just that. Oh and it repurposes the draft into tweets and a LinkedIn post.

The battle for attention on the internet has become more competitive overtime—we need any edge we can get. AI is advancements are rapidly increasing. The idea of a one-man media company is not that far fetched anymore. We can leverage it to do more in the same or less amount of time.

This essay explores the implications of the one-man media company as well as shows you the steps I took to build an AI Agent pipeline to move me closer towards becoming a one-man media company.

So let’s dive in.

Why is Media Important?

For centuries, media has been a pivotal force in shaping human discourse, once the domain of the few, now democratized by the internet's vast reach.

In today's landscape, media stands as a crucial lever among the triad that defines modern leverage, alongside capital and technology. It's an asset that amplifies ideas across the globe even as we sleep. The evolution from the printing press to television, and now to the internet, underscores a transformative journey towards accessibility.

With just a laptop and an internet connection, the power to broadcast messages and influence thought is unprecedented. This progression not only democratizes information but empowers anyone to become a central figure in the narrative of our time.

Media and Code to Multiply Leverage

Code, much like media, serves as a potent form of leverage in the digital age, operating tirelessly behind the scenes to amplify our efforts.

When we intertwine the vast reach of media with the precision and efficiency of code, we unlock a multiplier effect in leverage, propelling us into a realm where individual creators wield the power of entire media conglomerates.

This fusion births the concept of the one-man media company, where creators harness both content and technology to disseminate their ideas far and wide, with unparalleled efficiency. In this innovative paradigm, the synergy between code and media not only enhances productivity but redefines the landscape of content distribution.

Bye-bye New York Times

As code and content become easier to product for the individual, we will see more decentralized sources of information. The traditional media institutions don’t stand a chance.

The factors that make the one-man media company more feasible today are:

• Distrust in mainstream news—searching for direct sources.

• Advancements in AI—making content generation and automation easier.

• Fragmentation of interests into niche parts—more appetite for niche content.

These factors are wind in the sails of individual media sources. Now combine this with new ways to generate and automate content with code and AI. Boom. You’ve got yourself a one-man media business.

So now we know that media and code are a match made in heaven. Strong by themselves, but stronger together. Let’s now look at how I combined these two in my latest tool I prototyped this week.

Writers Room of AI Agents

Bridging the gap between the transformative potential of decentralized media and practical application, let's explore how the Writers Room of AI Agents harnesses this power, turning the one-man media company from a concept into a reality.

Demo

The structure of the pipeline is hierarchical with a top-level supervisor agent overseeing two sub-teams of agents: The Research Team and the Writing Team.

Now let’s break down the flow of this tool

1. Input Topic + Research

The process starts by having the user input a topic or idea they would like to write about. Then, the supervisor agent passes the input to our Research Team (which only consists of one agent right now). The search agent uses the Tavily API to search for relevant information, create a write up of the information and passes it off to the Writer Team.

2. Writing the draft

The information we get from the research is passed to our Writer team which then sends the information first to our Writer Agent. Once the writer agent has created a draft newsletter, we use our content repurposing agent to write out several tweets based on the article.

3. The Output

The output is saved to local files and is presented to the UI once the process has finished.

This flow has allowed me to save time I would have spent researching and drafting and reallocate that time towards editing and refining the idea.

The Stack

I’m currently using Streamlit, Langchain and OpenAI’s python SDK for this prototype.

If you’re interested in how I built this, here is a list of resources I found helpful:

1. Langchain’s Blog on Multi-Agent Frameworks here

2. OpenAI’s documentation here

Enhance not Replace

Like other tools I have made, this tool is not meant to replace but enhance us and our workflows.

I know you’re reading this and thinking “Why the hell does this guy want humans doing less work?”—that’s the wrong question to be asking; instead you should ask “What can I do now with this newfound time?!”.

The goal is to find ways to use AI that give us more leverage with our time. As AI continues to improve, we can delegate more time-consuming tasks to it while we focus more of our attention on creativity and learning.

Closing Thoughts

As we enter a new era of digital content creation, it’s clear that AI is going to be a massive point of leverage for those looking to compete for attention.

The journey from a single sentence to a comprehensive draft, further repurposed across various platforms, underscores a revolutionary shift in how we approach media production. This evolution towards a one-man media company, powered by AI, does not diminish the value of human touch in content creation; instead, it amplifies it.

By leveraging AI, we're not just streamlining processes; we're redefining the very essence of creativity, allowing for a level of productivity and engagement that was once the exclusive domain of larger entities.