Register for the Next Webinar (We'll notify you via email)
This Tip of the Week covers topics including:
- 00:00 - Website Content with AI Workflows (n8n + BD automations) (Intro)
- 00:55 - Meet Scott Germaise
- 02:55 - What We’ll Cover
- 03:14 - How Scott Automates BD with n8n (Demo)
How to Solve the Cold Start Problem with n8n + Brilliant Directories
Every new directory site faces the same trap: you can't market a site with no listings, and you can't get listings without first doing the work that makes the site worth visiting. In this Tip of the Week, Scott Germaise — a tech veteran with multiple exits and a background spanning Prodigy, About.com, e-commerce, health tech, AI, and web3 — shows how he used the AI workflow tool n8n to break out of that trap on his own Brilliant Directories site in roughly six weeks, using 40 to 100 hours of personal time to populate several thousand listings.
This article walks through the who, what, where, why, and how of Scott's setup, plus the benefits, real-world use cases, and how to start replicating it without writing code.
Who: Meet Scott Germaise
Scott has been building on the internet since its early days. He stumbled onto Brilliant Directories while researching decentralized technology, found the platform easy to work with, and decided to build a directory for the web3 / stablecoin / crypto-project space. The site had been live for about six weeks at the time of this session — a perfect case study for the cold start problem, because Scott was solving it in real time.
What: An AI-Powered Workflow That Builds Listings For You
The system Scott demoed has two main workflows:
- Listing builder — a workflow that takes a list of URLs from a Google Sheet, researches each company with an AI agent, categorizes the result, deduplicates against a Supabase database, and pushes finished listings into Brilliant Directories via API.
- Weekly news roundup — a Sunday-night trigger that aggregates RSS feeds from sources in his niche, formats the content into HTML, and auto-publishes a blog post on his BD site — keeping the content calendar fresh while sending traffic back to the original authors.
n8n is the connecting tissue. Unlike Zapier or Pabbly, n8n lets the AI itself build and modify the workflow nodes — you describe what you want and it produces the wiring.
Where: The Stack
- n8n — AI workflow automation (the orchestration layer)
- Google Sheets — input list of company URLs and working scratch pad
- Supabase — optional external database for canonical reference and dedupe (you can skip it and push directly to BD)
- AI agent — Grok, ChatGPT, Anthropic Claude, or any other model n8n supports (swappable)
- Botster — screenshot tool for capturing company website hero images
- Brilliant Directories API — the publishing destination
Why: Four Problems This Workflow Solves
- The cold start problem. A directory with 10 listings looks abandoned. A directory with 3,000 looks credible — and now you have something worth marketing.
- Manual research is brutally slow. Looking up a company, writing a description, categorizing it, and uploading takes 10–15 minutes per entry. AI agents do it in seconds.
- Content freshness for SEO. Google (and increasingly AI agents) reward sites that publish regularly. A weekly auto-generated news roundup keeps the blog active without staff.
- Replicability across niches. Once one workflow works, the same scaffold runs on any vertical — legal, medical, trades, finance — with just the prompt and source list swapped.
How It Works: Inside the Listing-Builder Workflow
1. Input: A Google Sheet of Company URLs
Scott maintains a Google Sheet as the working queue. New URLs go in, and the workflow handles the rest. Anyone on his team (or contractors) can add rows without touching n8n.
2. Dedupe Against Supabase
Before any AI calls (which cost tokens), n8n checks whether the listing already exists in his Supabase database. Supabase is optional — for smaller sites you can skip this and let BD itself be the canonical store — but it's useful when you're building thousands of records and need a separate "source of truth."
3. AI-Driven Company Research
For each new URL, n8n passes the site to an AI agent with a system prompt: "You are a web3 business directory researcher. Given this company URL, return a JSON object with these keys…" The model researches the company and returns structured data — company name, canonical URL, top-level category, subcategories, full description, short search description, and so on. The agent is swappable: Grok today, Claude tomorrow.
If you want to go deeper on the AI side, 5 Claude & Cursor Prompts to Grow Your Membership or Directory Site covers prompt patterns that pair well with n8n's agent nodes.
4. Manual Sanity Check
This is the deliberate human-in-the-loop step. The AI will occasionally hallucinate or miscategorize, so Scott reviews the spreadsheet output before pushing to BD. Quality control beats raw speed.
5. Image Capture with Botster
Botster screenshots the front page of each company's website. Currently a manual trigger, but the API allows for full automation in a future version of the workflow.
6. Push to Brilliant Directories via API
The final node sends the cleaned, categorized data through the BD API using an import ID and an API key. Images can be auto-fetched by the API itself. Scott receives an email notification when the batch completes, and the listings appear in his BD admin panel for one last review. For a primer on connecting AI tooling to BD, see Connect Claude Code to Brilliant Directories.
How It Works: The Weekly News Roundup Workflow
- Trigger: a scheduled node fires every Sunday night.
- Aggregation: a code node fetches and parses RSS feeds from sources in his niche. The code itself was written by an AI (ChatGPT/Claude) — no programming required from Scott.
- Formatting: the aggregated items are built into HTML (also AI-generated).
- Publish: the HTML posts directly to BD as a blog article, crediting original authors and linking back.
- Pacing: Scott interleaves original posts (some AI-co-piloted, some by contractors) between the auto-roundups so the blog feels editorial, not robotic.
The combined effect: a steady drumbeat of fresh content that powers SEO without dominating his calendar.
Benefits at a Glance
- Speed: thousands of listings in weeks, not years
- Cost: 40–100 hours of personal time vs. months of full-time research
- Quality control: manual review steps where they matter, automation everywhere else
- Model flexibility: swap AI providers without rebuilding the workflow
- SEO compounding: auto-published news keeps the site fresh and link-worthy
- Niche-agnostic: the same workflow scaffold works for any vertical
- Open source friendly: Scott shares his workflows as JSON on GitHub so anyone can import and adapt them
- No coding required: AI generates the code nodes themselves
Use Cases
- Brand-new directories facing the cold start problem — populate enough listings to look credible before marketing starts.
- Niche operators in industries where company data is publicly available (web3, SaaS, legal, medical, real estate).
- Multi-site portfolio operators who want one repeatable methodology across multiple verticals. Pair with a 90-day growth plan to scale fast.
- Content marketers who want a weekly automated news roundup that compounds SEO without staffing a writer.
- Operators monetizing through volume. Once the listing count is high enough, the directory becomes a canonical source — opening up monetization strategies like featured listings, paid placements, and sponsorships.
How to Get Started
- Pick your niche and source list. Scott started with a Google Sheet of web3/stablecoin/crypto URLs — yours might be law firms, gyms, or contractors in a region.
- Generate a BD API key from the Developer section of your dashboard.
- Set up n8n. Either self-host or use n8n cloud.
- Build the listing-builder workflow first. Start with one trigger, one AI research node, one BD POST node. Ship 10 listings end-to-end before expanding.
- Add dedupe and review gates only once the basic pipeline works. Premature complexity is the #1 killer of n8n projects.
- Layer in the news-roundup workflow once your listing base is large enough that the blog feels useful.
- Tighten the join experience on the front end so the traffic your fresh content attracts converts — see Build a High-Converting Join Page.
- Pair with deeper AI SEO. For ongoing optimization once the cold start is solved, see Claude Code for Online Directories + Membership Sites: AI SEO on Autopilot.
Watch the full demo above to see Scott's actual n8n canvas, the JSON the AI agent returns, and the Brilliant Directories admin panel populating in real time.
AI-Generated Transcript – Please excuse any inaccuracies
Website Content with AI Workflows (n8n + BD automations) (Intro) (00:00:00)
- Scott Germes is a special guest with a lot of experience in tech and is familiar with creating workflows connecting different programs and applications together (00:00:08).
- The topic of discussion is about automating website content using N8n, which is a leading market tool that allows users to create workflows by telling the AI what to do, without requiring manual mapping (00:00:29).
- n8n is an AI-powered tool that builds workflows with the user, and Scott has mastered using it with his Brilliant Directories site and other applications (00:00:43).
- The conversation is expected to cover how Scott has used n8n to automate his website content and other applications, and will likely provide tips and tricks on how to do so (00:00:49).
- The discussion is also related to other integration tools like Zapier or Pabi, which are used to create workflows connecting different programs and applications together (00:00:22).
Meet Scott Germaise (00:00:55)
- Scott Germaise has a background mostly in tech, having worked on the internet since its early days, and has been involved in various industries including e-commerce, health tech, AI, and web 3, which led him to Brilliant Directories (00:01:18).
- He has experience with multiple exits, including working with companies like Prodigy and about.com, and has been in the tech industry for a long time, giving him a unique perspective on the field (00:01:24).
- Scott Germaise found Brilliant Directories by accident while looking for a particular type of content related to web 3 decentralized technology, and decided to build a directory using the platform, which he found easy to use (00:01:46).
- He started putting content up on the site, which is still an experiment for him, and has been live for only about a month and a half, posing the challenge of the "cold start problem" and finding ways to build up the site (00:02:06).
- To address this challenge, Scott Germaise decided to use N8n, a workflow or automation tool, to help automate certain tasks and build up the site, and will be sharing his screen to show how he uses the tool with Brilliant Directories (00:02:26).
What We’ll Cover (00:02:56)
- The focus of the discussion is on a site referred to as Webpoint X, which will be the central point of exploration (00:02:56).
- The conversation will cover how Brilliant Directories listings are generated, as well as the process of determining categories and other related topics (00:03:03).
- The presentation will involve sharing the screen to take the audience on a step-by-step journey of how the process is being implemented (00:03:08).
How Scott Automates BD with n8n (Demo) (00:03:15)
- The website in question is a directory for decentralized technology, including stable points and crypto projects, and it features listings, upcoming events, blog posts, and news articles, with the goal of providing a comprehensive resource for users (00:03:15).
- The website was built using a tool called N8n, which is a workflow automation tool that takes a list of URLs from a Google sheet, researches the companies, and automatically builds a listing, categorizing it and updating the spreadsheet accordingly (00:04:06).
- The n8n tool uses AI to enrich the listings, and in this case, it is using a combination of Superbase, an external database, to keep track of the listings and provide a canonical reference list (00:05:33).
- The workflow automation process involves checking if a listing already exists, and if not, it uses an AI agent, such as Grock, to research the company and build the listing, following a system prompt to gather the necessary information (00:06:18).
- The AI agent can be replaced with other options, such as ChatGPT or Anthropic, and the n8n tool allows for customization and flexibility in the workflow automation process (00:06:25).
- The end result of the workflow automation process is a comprehensive and detailed listing, including a search description, about me section, and other relevant information, which can be used to import data into Brilliant Directories via API (00:05:06).
- The process involves taking inputs from an earlier node, researching a company, and outputting certain results, with a larger message providing context for the task, including a prompt that instructs a web 3 business directory researcher to research a company given a website URL and return a JSON object with specific keys (00:06:53).
- The system is populating a spreadsheet by researching companies and generating content, with the option to use a GPT or AI tool to assist with the process, and it is also categorizing listings and checking for existing entries in the database (00:07:06).
- The output of the node includes the researched company information, and the system is also capable of handling errors, such as those made by the AI, which can be reviewed and corrected manually (00:08:20).
- After the automated research is complete, a manual step is taken to review the results and ensure they make sense, and then an image capture tool called Botster is used to capture the front page of each company's website, which can be done manually or potentially automated using an API (00:08:51).
- The image capture process involves manually putting in URLs and capturing the front page of each website, and then manually adding the images to a sheet, with the option to use an API to automate this step in the future (00:09:07).
- The final step involves loading the researched and categorized company information into Brilliant Directory, which is a manually triggered process that could be automated further, but is currently done manually to ensure quality control (00:09:42).
- The process of loading the data into Brilliant Directory involves setting an import ID, using a node to retrieve the Google sheet, and looping over the different items in the sheet to prepare the listing data, with the option to automate these steps using an API (00:10:09).
- The system is designed to automate the process of gathering and uploading company information to a directory, starting with a list of websites, and it sends an email notification when the process is complete, allowing for a large number of entries to be processed at once (00:10:28).
- The process involves researching the companies, checking for duplicates, determining the top category, and merging results, before uploading the information to Brilliant Directories using an API key (00:10:41).
- The API key is used to automatically upload the company information to Brilliant Directories, and the process can be executed quickly using a workflow, which can be monitored in the admin panel (00:11:13).
- The automated process can also import images for the companies, either by manually selecting and fetching the images or by using an auto-import parameter in the API (00:12:20).
- The system generates a range of information for each company, including the company name, canonical URL, top-level category, subcategories, description, and short search description, before uploading the data to a database and Brilliant Directories (00:13:08).
- The automation process allows for efficient management of a large number of company listings, and the use of a database, such as Superbase, is optional, as the data can be uploaded directly to Brilliant Directories (00:13:49).
- The value of directory websites, such as the one being automated, lies in their ability to become a centralized hub of information, providing a useful resource for users, as noted in comparison to another user, Gerald (00:14:13).
- The process of creating a directory for a specific niche involves gathering information from various sources on the internet, and this information can be used to create a canonical source and source of truth for the industry, which can then be monetized, with the help of AI agents that can speed up the process (00:14:21).
- Having one project can lead to the creation of multiple projects and sites using the same methodologies, allowing individuals to work on various industries and verticals, and AI agents can assist in this process by helping to create content, such as listings and news roundups, more quickly (00:14:57).
- The use of AI agents can help resolve the "chicken or egg" scenario by allowing individuals to get content on their site quickly, making it possible to start marketing efforts with a site that appears to have been online for a while, and this can be achieved through the use of workflow automation tools like N8n (00:15:14).
- The n8n workflow automation tool can be used to gather information and load it automatically, allowing individuals to build several thousand listings in a short period, such as a couple of months, with a relatively small amount of personal time invested, around 40 to 100 hours (00:15:51).
- Creating a blog section on a site can involve writing original content, using AI as a co-pilot, and fact-checking, with some posts being written by contractors, and this can help establish the site as a credible source of information, with the use of AI to assist in the writing process (00:16:32).
- Automating certain tasks, such as a weekly news roundup, can be achieved through the use of AI and workflow automation tools, allowing for the creation of fresh content that sends traffic to the original sources, and this can be done in a way that credits the original authors and provides value to the readers (00:17:10).
- A workflow in N8N is used to aggregate content from various websites using RSS, which is a format for blog posts, and the process is automated using a trigger node that runs on Sunday nights (00:17:48).
- The trigger node uses code to fetch and parse news feeds from specified sites, and this code can be built with the help of AI tools like ChatGPT or Claude, which can generate the necessary code to be used in the N8N node (00:18:05).
- The aggregated content is then built into HTML, which can also be generated using AI tools, and the resulting HTML is posted directly to Brilliant Directories, keeping the content fresh and updated (00:18:54).
- To keep the blog looking active, original content is also posted, and it is tried to be dated to be interstitial with the weekly updates, so it doesn't look like only weekly updates are being posted (00:19:38).
- The workflows used in N8n are shared as JSON files on GitHub, and they can be imported into N8N, allowing others to use and modify them for their own purposes, although they may need to be tweaked to work with personal databases and APIs (00:20:05).
- To use the workflows, it is not necessary to be a programmer, as AI tools can be used to generate code and provide instructions on how to hook up nodes to APIs like Google Docs Sheets API (00:20:41).
- The entire process of building and automating the website content is done using N8N and AI tools, and it can be replicated by others with relatively little skills and knowledge, as demonstrated by the fact that the website was built in just a month and a half (00:21:03).
