How to create your first AI agent to track your competitors with Gumloop? 🕵️‍♀️

EDITOR’S NOTE

Dear future-proof humans,

A few weeks ago, I attended a Gumloop workshop, built a tiny agent, and felt that click when a boring task becomes simple. We wired five blocks, searched, scraped, extracted, filtered, and emailed, and it started sending clean competitor updates on its own.

I remember how we used to do this. Python scripts, cron, brittle scrapers, and one engineering teammate who knew the whole thing. Today, you can drag nodes in Make, n8n, or Gumloop, hit run, and ship in an hour.

Why do these visual builders matter now?

1. Speed for small teams: You can ship a working monitor in a single sitting, no tickets or waiting on a dev sprint.

2. Clear handoffs: Nodes read like a checklist, so marketing, product, and data teams can jump in and edit without fear.

3. Proof that the bet is sound: The wider automation market keeps growing, and adoption of low-code and no-code tools is rising across teams.

What you will find inside this issue of Nanobits: A simple plan to build a competitor monitor with Gumloop, using five nodes.

Let’s get you a monitor that works while you sleep so that you can focus on the next big thing.

WHAT IS GUMLOOP AND HOW DOES IT WORK?

Gumloop is a no-code AI automation tool. You build “flows” by dragging nodes onto a canvas and then connecting them to pass data from one step to the next, no scripting required.

The tool works best for web tasks. It can scrape sites, extract the parts you care about with AI, then route the results to email, Sheets, Slack, or your CRM.

Flows start with a trigger. Time-based schedules, a new email, a new row in Sheets, a webhook, or a Slack message can kick things off. You can watch runs, handle errors, and keep the flow on a set schedule.

You are not locked to one model. Gumloop supports OpenAI, Claude, Gemini, Llama, and more, so you can pick the right model for each step.

Security is production-grade. The company lists SOC 2 Type II and GDPR compliance in its Trust Center and docs.

Getting started is simple. There is a free plan with monthly credits, and you can scale seats and usage as you grow.

Typical build, end-to-end

1. Pick a trigger, for example, every morning at 9 AM.

2. Add a scraper node to fetch target pages.

3. Drop an AI node to extract, label, and summarize the results.

4. Send the output to your inbox or a team channel, then set the schedule.

That is the core idea: small blocks, clear flow, and quick results.

BUILD YOUR FIRST WORKFLOW WITH GUMLOOP

First, we learn what an AI agent is. While you may come across various definitions of it on the internet, here’s how we define it: An AI agent is any workflow that incorporates artificial intelligence. Without AI, you're simply building basic automation. Every AI-powered workflow follows a three-part structure: trigger, AI action, and results. The trigger starts the workflow, the AI performs actions based on the information you provide, and finally produces results from those actions.

Let's build a simple competitor tracker agent to see this in action.

Create an account:

• Create an account with your email or Google account.

• You will get 2,000 free credits to get started.

• If you watch their explainer video, complete their product tour, and create a sample agent, you'll receive an additional 1,000 credits for free. This should be enough for at least two flows.

Create a new workflow.

Click on “create workflow“ in the top right corner in the “Personal Workspace“ section of the tool.

Step 1: Add a Web Search node

For this experiment, we will consider OpenAI as our competitor. You may choose any one of your competitors. This node acts like a focused Google search for OpenAI updates on openai.com. It ensures that you always pull the most relevant updates directly from the source.

1.1 Add Web Search node

• Open the left panel.

• Find “Web Search”.

• Drag it onto the canvas.

1.2 Configure your search

• Click the Web Search block.

• Turn Use Advanced Search to True.

• Set Query to:

"OpenAI" product updates OR new features OR announcements site:openai.com

• Set Results Count to 5 (you may choose a higher number; this will impact the number of credits used for this operation).

What this does
The node searches openai.com for product updates, new features, and announcements. The quotes around “OpenAI” force exact matches, so fewer random results appear.

Step 2: Website Scraper

Web Search returns titles and snippets. The scraper opens each link and reads full pages. It gives you the actual article text, not just a preview.

2.1 Add Website Scraper

• Open the left panel.

• Drag “Website Scraper” to the canvas.

• Place it next to your Web Search node.

2.2 Connect the nodes

• Drag from Web Search to Website Scraper to create a line.

• This passes the found URLs to the scraper.

2.3 Configure the scraper

• Click “Website Scraper”.

• Set Use Advanced Scraping to True.

• Confirm the URL field maps from Web Search results.

What this does
The scraper visits each URL from Step 1 and extracts the text of the corresponding page. You get full content, not just headlines.

Step 3: Extract Data

Raw website content is messy and long. This step cleans it into short, usable notes. It saves you from scrolling through paragraphs to find one or two updates that matter.

3.1 Add Extract Data node

• Find “Extract Data” under Using AI.

• Drag it to your canvas.

• Connect Website Scraper to Extract Data.

3.2 Set up data fields

• Open the Extract Data block.

• Add two fields:

  • article_summary: Brief summary of the article.

  • key_points: Main points or announcements.

3.3 Additional settings

• Keep Extract List? set to No.

Step 4: AI Filter

Not every page on openai.com counts as news. This step keeps only useful updates. It saves you from reading blog fluff or minor changes that don’t matter.

4.1 Add AI Filter node

• Drag the AI Filter to your canvas.

• Connect Extract Data to AI Filter.

4.2 Set up the filter logic

• Open the AI Filter block.

• Set Filter By to article_summary from Extract Data.

• Set Value to article_summary from Extract Data.

• Set Condition to:

Filter for content related to product updates, new features, pricing changes, partnerships, strategic announcements, or major company developments. Include items about new AI models, API updates, enterprise features, or significant business moves. Exclude general company news, job postings, marketing content, or minor updates.

• Choose Claude 4.1 Opus as the model for best accuracy. The model you choose also determines the amount of credits that will be consumed during this operation.

What this does
The filter checks each summary and passes only real updates. You get fewer false alerts and a cleaner feed.

Step 5: Gmail Sender

This is your delivery system. It packages the important updates and sends them to your inbox. It makes sure you never miss a key change, even on busy days.

5.1 Add Gmail Sender node

• Drag Gmail Sender from the Gmail section.

• Connect the AI Filter to the Gmail Sender.

5.2 Set up your email

• Recipient: Enter your email address.

• Subject: 🔍 Competitor Update: OpenAI News Alert

• Sender Display Name: Gumloop Competitor Agent (or any name you prefer).

• Body: Map to filtered_output from AI Filter.

• Credentials: Connect your Gmail account, and Gumloop walks you through the steps. The process is easier on Gumloop than on make.com or n8n.io.

What this does
It sends a clean, professional email with only the important OpenAI updates, formatted and ready to read.

Here’s what happens when the agent runs:

• Web Search finds OpenAI pages that match your keywords.

• The Website Scraper opens each link and extracts the full text.

• Extract Data writes a summary and key points.

• AI Filter keeps only the updates that matter.

• Gmail Sender sends the final email to your inbox.

Time to test your agent

Step 1: Save everything

• Click Save at the top.

• Name it OpenAI Monitor or any other name that you prefer.

Step 2: Run your first test

• Click Run in the top right, the play icon.

• Watch each node light up as it processes.

• Look for any red error messages.

Step 3: Check your results

• Open your email and find the alert.

• Review what it found on the pages.

• Adjust the search terms or filter rules, then run again.

Congratulations! 🎉 You just built a working AI competitor monitor. It tracks OpenAI updates, filters noise, and emails clean summaries. You can build your business while your agent keeps watch.

Next steps: duplicate this for Anthropic or Grok. Or change the query to track pricing changes or enterprise features.

GUMLOOP vs N8N vs MAKE: CHOOSING THE AUTOMATION PATH THAT’S RIGHT FOR YOU

The automation landscape has evolved dramatically, with three distinct platforms emerging as leaders in the no-code and low-code workflow automation space. Gumloop represents the AI-first approach, Make.com offers user-friendly visual automation, and N8N.io provides developer-centric flexibility. Each serves different user needs and technical requirements, making the choice highly dependent on your team's capabilities and automation goals.

Understanding these differences is crucial as automation becomes central to business operations. The right platform can accelerate productivity, while the wrong choice can lead to limitations, unexpected costs, and technical debt. This comparison examines the key factors that should guide your decision.

The choice between these platforms ultimately depends on your team's technical expertise, specific use cases, and organizational requirements. Gumloop excels for AI-native workflows, Make.com provides the best balance of power and usability, while N8N.io offers maximum flexibility for technical teams willing to invest in setup and maintenance.

END NOTE

That is a wrap for this week!

You built a working competitor monitor in Gumloop. We set up Web Search, Website Scraper, Extract Data, AI Filter, and Gmail Sender. You saved the flow, ran a test, fixed any errors, and tuned the filter. Now updates land in your inbox without extra work.

We also compared Gumloop, n8n, and Make. Pick Gumloop for fast web monitoring with AI blocks. Pick n8n for control and self-hosting. Pick Make for a polished cloud tool with an extensive app catalog.

Next steps are simple. Duplicate this agent for Anthropic or Google AI. Or switch the query to track pricing changes and enterprise features. You can also log results to Sheets and post alerts to Slack.

Here are a few things you can try on your own that will spruce up this agent and make it more efficient:

1. Create a Pricing and policy watcher. Track pricing pages and ToS pages, highlight changed lines, and flag dates.

2. You can also build a Docs and API radar. Watch changelogs and model pages, filter for new endpoints, rate limits, or SDK notes.

3. Next, you can add a Partnership and funding module. Monitor press rooms and news pages for alliances, launches, and raises.

4. Then check for hiring signals. Parse job posts for team size, location, and stack, then tag growth areas.

5. Lastly, you can add an impact score by using AI to categorize these updates as high, medium, or low, and send a weekly digest to your preferred email, Slack, or Excel sheet.

If you found this exercise helpful, subscribe to our newsletter for the next build. In the coming week, we will try an email triage agent that categorizes mail as high, medium, or low priority. Until then, try creating this agent and start spying on your competitors.

Pro tips for success

• Start simple. Run the flow a few times manually and see what it finds.

• Tune the filter. If emails feel noisy or too sparse, adjust the criteria.

• Track more rivals. Duplicate the flow for Anthropic, Google AI, and other similar services.

• Set expectations. Expect 1 to 3 emails per week, depending on OpenAI activity.

• If you get stuck, reply to this email or reach out to us on LinkedIn and we'll walk you through the setup.