BuildFlows n8n Nodes

[00:02] All right. Today we are going to be reviewing over some of the things I have created on n8n. It's quite exciting. So I will dive right into it. And this is an example workflow I put together just to showcase and outline. I'll start it so we and then we can start talking through it here. But essentially what this workflow is doing is it will go through and gather all of the companies and the projects that this user is associated with. It will loop over each project and gather necessary data. From there, I just am filtering it down to one project for this example to then feed that into the AI agent. Filtering it down just to make it quicker for this example to talk through. But what I will do is do some back and forth between the chat and leveraging the tools that we

[00:47] have created and it will output a specific report that we have then requested. So it'll take a moment to run. Once it's done, it will complete. what we'll do then next is I will dive into the specific nodes that I've created. So the first one being Procore. So any endpoint that is available in Procore I have created. So there is 1,400 or so action these are different types of anything you can do in Procore you can basically do in here as well as then I also created some triggers. These are things like change events or budget changes or RFI created or any type of updates within the system. And so these can be

[01:32] leveraged to be able to interact with the Procore instance. I also put together a Autodesk node. So this Autodesk node outlines around 300 or so actions. Any endpoint also any endpoint that's available in Autodesk is mapped out here. And I put together a few triggers just for project creation. Couple of the other nodes that I have created specifically are the HCSS. So I just started building out this one has around 48 or so actions. I'll continue to progress along that. And further another great one that is super helpful is the Power BI. So Power BI

[02:17] specifically is being leveraged to be able to interact with different gateways, data sets, data sources. You can leverage groups and or workspaces that people are associated with to interact with these and various types of ability to manipulate the data sets via tables, rows, columns and so yeah, great value add here in terms of the nodes. then specifically when we get into the MCP servers, I have the MCP for Procore created here. So let me close out of there and dive into this MCP. So MCP is running on my localhost 8000 over SSE. I didn't include any authentication for this. the authentication is currently handled in

[03:03] the back end right now. And I'm selecting our different tools that we can give our agent access to. All of these tools I mapped out for any endpoint that is available in the Procore financials. So this is specifically only leveraging the get requests and or like list requests. but the idea here is that initially when people are first starting they don't want to have an AI update or make any changes in their system at this point. soon they will but the idea here is let's first have people leveraging it to allow for the AI to gather data to get data and then perform some form of analysis from that and so the first initial iteration which is exciting

[03:48] to get moving so this is the MCP server. Another important thing to touch upon is that within each of these node request what we're doing is I added the authentication method so in the credentials we're leveraging the Procore authentication method through the client ID and client secret. But as you'll see here, I specifically have created a build flows API key. What this API key does is it will actually authenticate to ensure that that user has access to that node. And then what we can do is leverage that to be able to then do an analysis on our side to ensure we're updating these nodes properly that we're able to identify various information and then use that information to whether

[04:33] create specific AI agents or fine-tune agents or the large language models on their data and or doing any type of RAG for identification. So diving into that you'll see here I'm specifically in our I'll go back into the dashboard here but in this dashboard we'll get a general overview we can create a new customer to provision the API key and we can even go through and deploy a new droplet instantaneously utilizing DigitalOcean this will spin up their own instance of n8n and automatically install our pre-installed nodes that we have created here when I go into the details tab this will then go and show this specific company, the products that they have access to, if

[05:18] they have any deployments, and a deep analysis of how they're functioning and if there are any failed requests. And so the cool thing being is that we can then identify each flow run, if it's been successful, how many nodes are within it, and how many requests then for each of those nodes. And so we can then start getting granular in terms of identifying specific information whether it's the input requests or the response that is associated with that. So giving us the ability to do deeper analysis of if API endpoints need to be updated and or if we want to leverage this data to be able to go through and fine-tune a large language model for that company specifically and or if we want to go

[06:03] through and do some type of RAG so people can then search and or query for their specific data. But this is important to be able to do any type of change detection and analysis for the future. So that's what I have created. let me know if you have any questions and if you have any specific requests eager to develop and grow further and so thank you very much and I appreciate the time.