How to Start an Online Business with AI Agents No Coding

You have probably heard people talking about AI agents for the past year and felt completely left behind. Everyone assumes you know what an API key is. Everyone assumes you have some kind of technical background. And if you do not, you are sitting there watching video after video, hearing about all of these great opportunities, and still not knowing where to begin. That is a real problem. And it is the exact problem this walkthrough exists to solve.

In this video, Alston walks through the entire process of building your first AI agent from scratch using N8N and OpenAI, no coding required. Not a high-level overview. Not a theory video. An actual step-by-step build where he shows you every click, every credential setup, and every connection. By the end, you have a working agent that pulls trending topics from Google Trends every morning at 6 a.m., generates niche ideas with monetization suggestions, and drops everything into a Google Doc automatically. That is not a demo. That is a business tool you can run starting today.

What You’ll Walk Out With

A plain-English understanding of what an AI agent actually is and what it does
A free N8N account and a working workflow canvas
Google Docs connected to N8N via OAuth 2 credentials from Google Cloud Console
An OpenAI API key wired into your workflow so ChatGPT does the thinking
A niche-finder agent that runs daily at 6 a.m. and feeds you fresh content ideas every morning
The modularization mindset: why one agent, one task is the right way to build, and why chaining 15 steps together is how beginners burn hours debugging
A map of what to build next once this first agent is running
Clarity on which online business direction fits your situation, get that at finder.platformproof.com

What an AI Agent Actually Is

Before touching any tools, Alston makes sure you understand the thing you are about to build. An AI agent, at its simplest, is an automated process. You give it a task, or a very specific set of tasks, and it executes them. It can learn from previous executions to get better. It works on your behalf, runs as often or as infrequently as you need, and shuts down when it is done.

That is it. No magic. No PhD required. Think of it like a virtual assistant who never sleeps, never forgets, and never needs to be reminded twice. You build the workflow once, and then it runs. The confusion comes because most people talking about AI agents assume you already understand concepts like APIs and credentials. This walkthrough assumes you know none of that, and builds from the ground up.

Inside an AI agent there is a workflow, and inside that workflow there is a series of steps. You can make those steps as detailed or as simple as you want. For your first agent, Alston keeps it very simple: one task, one output, one destination.

The Two Tools You Need to Get Started

There are dozens of workflow builders out there. Make.com. Zapier. Countless others. Alston uses N8N here, and there is a link to it in the video description. But the important thing he wants you to understand is this: do not fall in love with the tool. These platforms are interchangeable. What you are actually learning is how to build workflows and agents. The tool is just the canvas.

The second tool is OpenAI, meaning ChatGPT. You are using ChatGPT as the brain inside your agent. It receives directions through a prompt, and it generates the output. N8N is the plumbing that connects everything together. OpenAI is the intelligence that does the actual thinking.

Step 1: Create Your N8N Account

Go to n8n.io. When Alston first started using it, they offered a 7-day free trial. By the time he filmed this walkthrough, it had expanded to a 14-day free trial. Either way, you do not need a credit card to get started: just a name, email, and password.

Once you are in, you will see the main homepage with a button that says Start Automating. Click it. Then click Start from Scratch. You are now looking at a blank workflow canvas, and the first thing you need to place is what N8N calls a trigger.

A trigger is the thing that starts the entire workflow. Think of it like an alarm clock. When it fires, everything else wakes up and begins running. Click the plus sign. Select Schedule from the list of trigger types. Set it to run every day, interval of one day, at 6:00 a.m. Click Execute Step to confirm it works. You will see a confirmation. That is your alarm clock, set and ready.

Step 2: Connect Google Docs with OAuth 2 Credentials

This is the step that trips most beginners up, so Alston walks through it in more detail than you will find in most tutorials. When you want to connect a popular service (Google Docs, Google Sheets, Dropbox, anything) to N8N, those two platforms do not automatically know about each other. You have to make the introduction. You have to get them to shake hands. The way you do that is through credentials.

In your workflow, click the plus sign and add a Google Docs node. Set the action to Create a Document. When you try to configure it, N8N will ask for a credential: specifically a Client ID and a Secret ID. These are like a username and password for the connection between N8N and Google. You would not share your bank password with anyone. Same rule applies here.

To get those credentials, go to Google Cloud Console in your browser. If you have a Google account, access is already included. Google also offers $300 in free credits for new users with a 90-day window to use them. Once inside Google Cloud Console:

Click on APIs and Services
Search for and enable the Google Docs API so that Google Cloud knows you want to allow a third party to interact with your Docs
Click Credentials, then Create Credentials, then select OAuth Client ID
Configure the consent screen: give the app a name (Alston calls his “test 1 2 3”), add your email, and click through the agreement steps
Select Web Application as the application type
Copy the OAuth redirect URL from N8N and paste it into the Authorized Redirect URIs field in Google Cloud
Click Create and you will get your Client ID and Secret ID

Copy those credentials immediately and paste them into N8N. Then click Sign In with Google. You may see a warning that says “N8N hasn’t been verified.” That is normal. Click through it, select all permissions, and continue. You will see Connection Successful. Google Docs and N8N are now shaking hands.

One important note: if you later want to connect other Google services (Sheets, Drive, anything else) you have to repeat this handshake process for each one separately. They do not share credentials automatically.

Now configure the Google Docs node to create a new document titled with today’s date. Drag the date field into the title. Set the folder to your home Google Drive folder. Click Create Test Step. If it works, go check Google Drive and you will see a blank document with today’s date as the title. Alston’s test showed “May 29th, 10:12:13.” That confirms the connection is live.

Step 3: Set Up Your OpenAI API Key

Back in your workflow, click the plus sign again and select AI from the node types. Then select AI Agent. Inside the agent, you need to connect a chat model. Select OpenAI Chat Model. N8N will now ask for an API key.

To get that key, go to the OpenAI developer platform. Search for “OpenAI API” in your browser. Create an account or log in if you already have one. Once inside the dashboard, look on the left side panel for API Keys. Click Create New Secret Key. Give it a name (Alston calls his “delete me” during the demo, since he planned to remove it after the video). Copy the key immediately.

This is critical: once you leave that page, the key is gone forever. OpenAI does not show it to you again. Save it somewhere secure before clicking away. Paste it into N8N, click Save, and you will see a success confirmation. Now select your ChatGPT model. Alston selects GPT-4, which he describes as “pretty standard” for this kind of workflow.

Step 4: Add the Google Trends RSS Feed as a Tool

Here is where the agent gets smart. Inside your AI agent node, there is a section for Tools. Tools are things the agent can use while it is running. Click the plus sign under Tools and search for RSS. Select the RSS Read Tool.

An RSS feed is essentially an approved web scraper. Certain websites publish RSS feeds so that third-party tools can read their latest content automatically. Google Trends publishes one. To get the URL, go to trends.google.com, click on Trending Now, and look for an Export option in the top right. Click RSS Feed. Copy the URL that appears. Paste it into your RSS Read Tool in N8N. That is the whole setup. Your agent can now read what is trending on Google in real time every single day.

Alston also adds a Think Tool to the agent. It is not mandatory, but he has been using it recently and it seems to improve output quality. You can add it the same way: search for it under Tools inside the AI agent node.

Step 5: Write the Prompt Using ChatGPT

Inside the AI agent node, set the prompt source to Define Below. This is where you tell the agent what to do. But here is a move Alston uses that most beginners do not think of: instead of writing the prompt yourself, go to ChatGPT and ask it to write the prompt for you.

Alston types something like this into ChatGPT: “You are an expert in prompt engineering. We are creating a prompt to help us build an AI agent that gives us niche ideas, subniches, and content ideas based on trending topics. Here is what we want for our agent…” and then pastes in his requirements. ChatGPT generates a full, structured prompt in response.

The prompt ChatGPT created for this workflow looks something like this: “You are a trend scout and expert AI agent in trend spotting, niche selection, ideation, and content strategy. Every time I invoke you, you will…” and then walks through each instruction. The agent is told to use the RSS feed and the Think Tool to ensure the best possible output. Paste the full prompt into the Define Below field in your AI agent node.

The agent is now instructed to do three things with every run:

Find trending topics by reading the Google Trends RSS feed
Generate niche and sub-niche ideas based on those trends, with monetization suggestions (affiliate marketing, digital products, or other)
Create a list of content ideas tied to each niche

Step 6: Connect the Output to Your Google Doc

The final step in the build is getting the agent’s output into the Google Doc you created earlier. Go back to your Google Docs node, but this time set the action to Update a Document instead of Create. You will need to pull in the document ID that was generated when you created the dated document in Step 2. Drag it into the Doc ID field. Then drag the AI agent’s output into the Text field.

Click Test Step. If it succeeds, go check your Google Doc. The agent’s output, formatted niche ideas, monetization suggestions, and content ideas, should now be sitting in that document. Alston’s test run came back with cruise travel (specifically Royal Caribbean as a trending brand), luxury cruise tips as a micro-niche, eco-friendly cruise travel as a second micro-niche, and women’s tennis apparel and tennis training for beginners as separate rising opportunities.

That output happened in minutes. The agent read Google Trends, processed the data through ChatGPT, and dropped the results into a Google Doc. Now activate the workflow and set it live. From this point forward, you get a fresh set of niche ideas and content suggestions delivered to that Google Doc every morning at 6:00 a.m. without touching anything.

Why the Output Actually Matters for Picking a Niche

One of the most common mistakes beginners make when starting an online business is not niching down far enough. Someone interested in tennis starts a channel about tennis. That is too broad and too competitive. The agent’s output helps you go deeper. Instead of “tennis,” you get “women’s tennis apparel” or “tennis training for beginners.” Those are real micro-niches with real audiences. They are specific enough to build a core following without going head-to-head with channels that have been running for a decade.

Alston’s point here is direct: there is no such thing as the perfect niche. If you can see yourself creating content in a space for six months, a year, six years, then that is the perfect niche. The agent does not pick it for you. It gives you options to react to, which cuts down the paralysis and the time you would otherwise spend staring at a blank screen.

Run the agent for seven days and you will have 70 different content ideas sitting in your Google Docs folder. Seven days of ideas, generated automatically, with no manual research required. That is the actual time value of this workflow.

The Modularization Principle: One Agent, One Task

Alston has a coding background, and one principle from software development shapes how he builds every agent: modularization. The idea is simple: do not make one agent do 15 different things. Build one agent for one task, let it complete that task, and shut it down.

The reason comes down to debugging. If your agent has 20 steps and breaks on step 14, you spend your entire day trying to figure out where it failed. If your agent has one step and breaks, you know exactly what went wrong in seconds. Keeping agents focused also makes them more reliable. They are less likely to get 10 steps in and lose the thread of what they were doing.

This is counterintuitive for a lot of beginners who want to build one giant agent that handles everything. Resist that. Build a niche-finder agent. Build a separate content-idea agent. Build a separate post-scheduling agent. Each one does its job. Each one is easy to test, easy to fix, and easy to replace if something changes.

Not sure which online business model fits what you already know?

Answer a few quick questions and get a clear direction at finder.platformproof.com.

What to Build After Your First Agent

Once you have this niche-finder agent running, you have the foundational knowledge to keep building. Alston walks through several natural next steps in the video:

Daily content idea agent: Now that you have a niche, build an agent that goes out every day and surfaces specific content ideas within that niche: not just broad topics, but angles, hooks, and formats.
Trend-tracking agent for YouTube or TikTok: Build an agent that scans what is performing well in your niche on those platforms and reports back. You see what is working and create your own versions of it.
Digital product idea agent: Use a similar workflow to brainstorm digital product concepts based on what your audience is searching for. This can directly inform what you create and sell.
Turn the skill into an agency: Once you can build agents reliably, you can help other people do the same thing. Beginners who want to start online businesses need exactly what you just learned. You can charge for that knowledge and that time.

The skill compounds. Each agent you build teaches you more about how nodes connect, how prompts work, and where the failure points tend to be. After five or six builds, you can put together a new workflow in under an hour. That is when the real payoff kicks in.

Honest Drawbacks

This walkthrough gives you a real starting point, but it is worth being honest about what it does not solve.

N8N is not free forever. The 14-day trial gets you started, but you will eventually need a paid plan to keep your workflows running automatically. Budget for that before you build anything you plan to rely on long-term.

OpenAI charges by usage. GPT-4 is more capable but costs more per token than cheaper models. If you run a daily agent that generates long outputs, monitor your usage dashboard so you are not surprised by the bill at the end of the month. GPT-3.5 works fine for simpler tasks and costs significantly less.

The agent gives you ideas, not a business. Niche ideas are a starting point, not a substitute for execution. The agent can tell you that women’s tennis apparel is trending. It cannot tell you whether you specifically will enjoy creating content in that space, or whether you have the patience to build a following over 12 to 18 months. That part is still on you.

Errors happen. During the live demo in this video, the agent returned an error on the first test run. Alston’s solution is exactly right: copy the error message and paste it into ChatGPT. Ask it what the error means and how to fix it. That is a faster and cheaper debugging path than anything else available to you as a beginner.

Google Cloud setup is the most confusing part. The OAuth 2 credential setup is not hard, but it has more steps than anything else in this walkthrough. If you get stuck, you almost certainly missed one of the redirect URI steps. Go back and make sure the N8N redirect URL is pasted exactly into the Authorized Redirect URIs field in Google Cloud. That is where 90 percent of the errors in this section come from.

Find Your X

AI agents can help you move faster once you know what you are building toward. The harder question, and the one most people stall on, is figuring out which online business model actually fits the skills, time, and resources you already have. finder.platformproof.com is a free tool that asks you a few targeted questions and gives you a clear starting point based on your actual situation. No generic advice. Just a direction that makes sense for where you are right now.

Frequently Asked Questions

Do I need to know how to code to build this agent?

No. The entire workflow in this tutorial is built through N8N’s visual drag-and-drop interface. You are connecting nodes, pasting credentials, and entering text prompts. None of that requires you to write a single line of code. Alston explicitly designed this walkthrough for people with zero technical background.

What does N8N cost after the free trial?

N8N has a free self-hosted option if you want to run it on your own server, and paid cloud plans for people who want the hosted version without managing infrastructure. The starter cloud plan typically runs a few dollars per month. Check n8n.io for current pricing since plans change. For most beginners, the free trial is enough time to build and test several workflows before committing.

Is GPT-4 required, or can I use a cheaper OpenAI model?

Alston uses GPT-4 in this walkthrough because it produces higher-quality niche and content ideas. But GPT-3.5-Turbo works for this kind of task and costs significantly less per API call. If you are just experimenting, start with GPT-3.5 and upgrade only if the output quality is not good enough for your purposes.

What happens to my OpenAI API key if I close the page before copying it?

It is gone. OpenAI shows you the key exactly once during creation. If you close the page without copying it, you will need to go back into your API Keys dashboard, delete the old key, and create a new one. That is not the end of the world. It just takes a few extra minutes. Get in the habit of copying the key and saving it in a password manager or a secure notes app immediately after creation.

Why use N8N instead of Zapier or Make.com?

All three platforms do similar things. Alston uses N8N in this video, but he specifically says not to fall in love with any one tool. Zapier is more beginner-friendly but limits the number of tasks on the free tier more aggressively. Make.com (formerly Integromat) is powerful and used heavily by agencies. N8N’s free trial and AI agent nodes made it the practical choice for this walkthrough. The workflow logic you learn transfers across all three platforms.

Can this agent pick my niche for me?

The agent gives you a list of trending niches and sub-niches with monetization angles attached. It does not know what you are interested in, what you are good at, or how long you are willing to show up for a topic before moving on. Alston’s advice in the video is direct: any niche is the right niche if you can see yourself creating content in it for six months, a year, six years. The agent cuts down your research time. The choice still belongs to you.

What is the RSS feed from Google Trends actually pulling?

Google Trends publishes an RSS feed of its “Trending Now” section, meaning the topics that are spiking in search volume at any given moment. When you go to trends.google.com, click Trending Now, and export as RSS, you get a URL that refreshes automatically as trends change. N8N’s RSS Read Tool fetches that URL each time the agent runs, giving the AI model a current snapshot of what people are actually searching for right now rather than working from static or outdated data.

What is the “Think Tool” and do I need it?

The Think Tool is an add-on inside N8N’s AI agent that gives the model a dedicated space to reason through the task before producing output. Alston describes it as something he has been using recently and finds helpful, but he is clear that it is not mandatory. If you are getting good output without it, skip it. If your agent is producing shallow or inconsistent results, add the Think Tool and see if the quality improves. Treat it as a dial you can turn, not a requirement.

Sources

Alston Godbolt, “How to Start an Online Business with AI Agents No Coding,” YouTube, Platform Proof channel (video ID: Vcc80IWDEIc)
N8N workflow builder: n8n.io
OpenAI API platform: platform.openai.com
Google Cloud Console: cloud.google.com/console
Google Trends Trending Now RSS export: trends.google.com

Helping 1 million working adults make their first $3,000 online with the skills they already have. Alston Godbolt, Platform Proof.