// transcript — 241 segments
0:00 Intro
0:03 this was my startup team in 2024 we were nine people in total but
0:10 this year we only four people left I replaced more than half of my
0:16 team with AI agents and in this video I'll show you how you can create your
0:23 own AI startup team yes you can do this without riding code every business is
0:28 going to have a um an AI agent that their customers can talk to agents will
0:32 begin to deliver on hype I got to tell you K opportunity for agents is gigantic
0:38 as we AI agents are AI systems that can do work for hundreds of millions of
0:41 billions of different AI agents billions of different AI agents whether you think
0:47 this is good or bad AI agents are here and you can be certain that they're here
0:52 to stay millions of jobs worldwide will soon be replaced by AI agents and
0:56 thousands of tech companies and small startups will reduce their headcount and
1:01 their dependency on human labor it's very likely that startup Founders just
1:05 like myself can go back to basics and become a solopreneur running everything
1:09 as a oneman business without reducing the output and without losing revenue
1:14 and for you the upcoming founder who wants to run a SAS business but is
1:19 blocked by the sheer thought of having to manage a team this is incredible news
1:24 we're still very early so don't miss out on this opportunity let's get ready cuz
1:29 we have work to do in this video I will show you step by step how you can create
1:35 a fine-tuned specialized AI model how you can use it as an AI agent and how to
1:40 make it collaborate with other AI agents on your team let's get started one area
1:46 of work I replaced with AI is content creation I used to have people help me
1:50 write content for social media for my weekly SAS newsletter and my YouTube
1:54 scripts so in this video we'll set up an AI agent that creates content for social
1:59 media though the process I'll demonstrate here can be used for almost
2:03 any type of task you typically run into on a startup team fortunately open AI
2:04 Using ChatGPT directly
2:09 has made this incredibly easy for us we can just ask chat GPT to generate some
2:14 content right hey chat GPT can you create a banger post about SAS and
2:21 Entrepreneurship for me please sure how about this exciting times in the world
2:26 of SAS and Entrepreneurship every day presents new opportunities to innovate
2:30 and lead in this dynamic industry uh that's no that's not something I would post uh try again
2:39 please okay how about this feeling grateful for the amazing journey of SAS
2:43 and Entrepreneurship here's to more coffee fueled brainstorms and chasing
2:49 dreams that's really tacky and lame and what's up with all the Emojis no all
2:55 right all right I got you leveraging and utilizing synergies in the existing
3:00 realm of SX you know what thank you this is not going to work so it turns out we
3:05 cannot just ask chat gbt to create content for us and expect it to create
3:09 anything of good quality and many people who try this will stop here they'll
3:15 conclude that AI is bad it's generic poor completely lacking substance and so
3:20 on but let's just back up for a second and compare this to hiring a new
3:24 employee If you hired a Content writer and you gave them no onboarding access
3:28 to no brand guidelines no initial instructions and just ask them to create
3:33 a post about some topic how do you think they do do you think you would magically
3:37 be happy with their first result or would you expect them to First learn
3:41 about your brand your core values your style of writing and the history of your
3:45 company and then start delivering good results cuz that's exactly what we need
3:51 to do to prepare our AI agent now I've seen many different approaches to this a
3:55 common method is including examples and instructions in the system prompt or the
4:00 first message of the conversation it rarely produces good results that's why
4:06 most of the custom gbts suck cuz that's exactly what they are it's just a chat
4:10 gbt with some instruction in a system prompt injected into the chat before the
4:14 conversation starts these are not AI agents even though some people
4:18 Fine-tuning an AI model
4:19 incorrectly call them that we need to fine tune a model in essence instead of explaining
4:27 what the AI should do we will show it a whole bunch of examples of what an out
4:32 outut should look like fine-tuning an AI model is the process of specializing an
4:37 AI by training it on specific data so in this case we will prepare examples of
4:42 social media content it can be your own content it can also be content from
4:46 other creators that you really like we should put together 20 to 50 good
4:52 examples now for each of them we need to reverse engineer The Prompt so take this
4:58 post and imagine chat GPT generated this what would the prompt have looked like
5:02 probably something like this perfect we need to repeat this for
5:08 all examples if you want to save some time you can actually ask chat gbt to
5:12 reverse engineer the promps for you in most cases it actually does quite
5:19 well now let's head over to platform. account click fine tuning in the left
5:30 side menu and click create this is where we create our fine-tuning
5:35 job but before we can upload our training data we need to turn it into
5:40 this specific format called Json L it's Json and it's one example per line if
5:45 you're a programmer you can write a quick script to automate this if you
5:50 don't know what Json is just follow my lead here for each example we need one
5:56 line on this specific format we have a field called messages which is a list
6:02 that has two messages one from the user that will be our [Music]
6:07 prompt and one from the assistant which AI if you want to include the line
6:23 breaks as part of the data you just need to use back sln there we
6:31 go once we have all our training data in it choose the base model I recommend
6:42 using gbt 40 and that's it we can leave everything else as is and click
6:51 create now open AI will start training model once it's ready we can go to the
7:07 Creating a brand brief
7:10 so let's try again create a post about SAS and Entrepreneurship okay let me see I was
7:15 failing SAS until I finally understood this your team should be remote you
7:20 should work async taking VC money is your best bet awesome that sounds exactly like my
7:29 way of writing love it wait a minute well what was that last line taking VC
7:35 money is your best bet no no we can't have that I'm all about bootst startups
7:40 so we can't have that thing with VC money ah okay got it got it how about
7:45 this one I was failing SAS until I finally understood this the easiest way
7:50 to create a bootstrap sass is using Cloud tools like versel or netfi no no
7:55 no absolutely not you sound like me okay but you're saying things I would never say and this
8:04 is our next problem now we have a fine- tuned model that has learned to adopt a
8:08 specific writing style but it's not really that valuable if it doesn't know
8:13 what our brand is all about it needs to know which core beliefs and opinions to
8:17 adopt which common themes can be found in our content which hot takes to drop
8:22 and which values to align with and this is not exactly something we can find you
8:26 in our way out of instead we're going to do exactly what we would do with regular
8:32 employees we should write a brand brief in plain English this brand brief should
8:37 provide clear writing guidelines describe Our Brands in terms of core
8:43 values content themes and objectives dislikes and opponents Associated
8:47 products and services and other areas of our brand write this in a text file
8:52 notion page a Google doc or whatever you prefer if you've never done this before
8:56 or you haven't really thought about your personal brand in this way I actually
9:00 think it's a great exercise to spend some time writing all of this down but
9:06 again if you want to save some time you can have chat GPT do this too simply
9:11 take all of the training data you used to find through your model in the
9:15 previous step upload it to chat GPT directly and ask it to create a brand
9:21 brief given these posts then edit it to your liking and simply use that for now
9:26 I've written my brand brief in text file and if we head back over to the
9:29 playground we can now upload this file Island ask or fine-tuned model to use
9:34 this brand brief as a base for creating our posts I have to say this is really good
9:36 Building a team of AI Agents
9:43 content no one likes a feature creature how did you come up with that well I
9:48 guess I learned from the best ah I think you will fit in very well on
9:54 this team though not going to lie it's a little bit timec consuming to have to
9:58 ask you to write every single post can't you work a little a little bit more
10:03 autonomously I would love to Simon but I'm just an AI model I can't do anything
10:08 on my own initiative or make any active decisions all I can do is answer you
10:13 when you ask me for something so far we've solved two major shortcomings of
10:19 the normal chat gbt we fine- tuned a model to write in a specific way and we
10:23 taught it about our brand to understand what we're all about it now knows what
10:28 to say and how to say it but it's still not an AI agent it doesn't do anything
10:33 on its own and it doesn't make any decisions to be fair at this stage is
10:38 still very useful I've been using this exact setup for a long time and it has
10:42 been tremendously helpful but it's also been very timec consuming chatting with
10:46 my fine tuned model refining giving it feedback copy-pasting the results into
10:51 feed hiive my social media tool and so on so at some point I got this idea what
10:57 if I created a small team of AIS and gave them the agency to collaborate with
11:01 other and push the final results into feed hiive all by themselves we can do
11:05 better how about an engaging question we should also use a more
11:09 consistent tone of voice here's how it would work we want one AI agent to generate topic
11:18 ideas we want another agent to create a post from one of these topics we want a
11:23 third agent to check the quality of the post and give feedback to the second
11:28 agent the second agent will use the feedback to refine and improve the
11:33 post these two agents will go back and forth for a bit until they agree that
11:37 the post aligns with the brand uses the right tone of voice and is of high
11:43 quality then finally push the post to feed High where it shows up under my
11:48 drafts let's set this up for this we need a tool called n8n n8n is an automation tool in the
11:57 same family as sapure and make but just way way better you can sign up and use
12:01 NN in the cloud but you can also install it locally or selfhosted which gives you
12:06 unlimited workflow runs totally free of charge let's set this up in nadn Click
12:08 n8n workflow: Get Brand Brief
12:12 create and then choose credential here you need to add the credentials to the apps you want nadn to
12:19 access such as notion Google Docs open Ai and so on you will at the very least
12:25 need to give nadn access to open AI by adding an API key there are many Guides
12:30 Online on how to do this so moving forward I will assume that you have set
12:35 up credentials for all the apps you need so now that we have that done let's
12:39 start cooking we'll start from the smallest task and work back up in the NN
12:46 dashboard create a new workflow let's brief add First Step since this is going
12:57 to be a subtask we will use when called by another workflow I have my brand
13:26 step there we go now because notion outputs a page in blocks we want to
13:31 aggregate at the parts we need so let's node what we're interested in is the
13:49 Fields there we go we now have a simple list of individual pieces of text so
13:59 node drag in our content list we want the final output to be one piece of text
14:05 so we'll change this to string and after done let's move on create
14:18 n8n workflow: Get Content Ideas
14:25 new call it get content ideas once again use when called by another workflow
14:32 now in order to get ideas we first need the brand brief so we'll use a call
14:36 another workflow step and we'll choose the brand brief created let's rename this to get brand
14:50 brief awesome next let's add a simple open AI note called messageer
14:57 model for this task we can simply choose in the text field give the model a
15:05 prompt that describes what it's supposed to do we'll tell it to create 10 topic
15:09 suggestions that aligns with the provided brand brief we'll tell it to
15:15 use a Json output format like this and we'll provide the brand brief from the
15:36 out that's perfect and also here let's add an edit note and let's map this over
15:42 to a simple list called suggestions like that let's move on to the next
15:47 n8n workflow: Get Content Feedback
15:51 workflow we call this get content feedback it's similar to the workflow we
15:55 just did so let's set up the execute workflow trigger and the get brand brief
15:59 notes just like before this workflow is a little different since it will take some
16:06 incoming data namely the post to give feedback on so let's double click on the
16:12 first node click set mock data then add this perfect and also here we want a node
16:23 from open AI that messages a model and also here we can choose gbt 4
16:29 mini this time we want to give the AI some rules to follow so let's change
16:34 this first message to the type system the role of this AI is to give
16:38 feedback and ensure the quality of a post is good so we will simply give it
16:43 some criteria to follow of course you should adjust this to fit your specific
16:51 brand the most important part here is to have the model give a score between Z
16:55 and one and show it in the Json format we want to use for the
17:00 output let's make sure to include the brand brief and this time we want to add
17:08 another message with the role user and this will simply be the query
17:12 from the first step also here we want to Output as it perfect and yeah since this is just a
17:27 mock input I'd expect the score to be pretty low here so that seems to work
17:31 finally also here let's make sure to bit there we go now we got everything we
17:43 need let's put the whole thing together I'll show you how to do this but if you
17:48 want to download and import all of these workflows directly into your own nadn I
17:52 will share access to these files with members of Founders dech I'll tell you
17:56 more about this towards the end of the video create a new workflow let's let's
17:57 n8n AI Agent workflow
18:00 call it content creator for now the first step will be on chat message but we'll change that in
18:14 [Music] agent now we need to give this AI agent three things a model memory and tools
18:27 for the model we'll use open AI [Music] and because we linked our open AI
18:32 account we have access to the fine-tuned model we created earlier for the memory let's just use
18:46 buffer now we need to give this agent some tools these tools are going to be
18:51 the workflows we already created so let's pick NAD and workflow let's give the tool a name get
18:57 brand brief let's describe list we'll add the feedback tool in the
19:14 way there we go awesome now let's double click the AI agent we need to give this agent a
19:22 description of its main objective so we'll use system message
19:28 and we'll give it a description similar to this a quick explanation of how we want it to
19:35 work how to use the tools we provided and which final output we
19:49 out all right and we can see that the models start collaborating going back and forth and
19:58 there we go look at that really cool at this point the basic AI agent
20:02 functionality is pretty much done the agent has access to a fine tun model it
20:06 has access to the tools it needs to access the brand brief and get feedback
20:10 and because it's an agent it figures out when and how to use these tools
20:15 correctly given its task description this right here is an actual AI agent
20:20 Connecting to FeedHive
20:21 let's finish up since we want the output of this AI agent to be passed onto
20:25 another tool in my case feed hiive We'll add an HTTP request node here
20:32 in feed hiive under automation we can create a new trigger I'll call this AI content and
20:41 pre-select a few social accounts I'll grab the URL from nadn we don't have images in this
20:54 workflow so I'll remove the media url's part and for the text I'll grab the
20:59 output from the AI agent agent and simply add it here now every time the
21:05 agent finishes a post it'll add it directly to feed hiive where I can
21:11 Putting it all together
21:13 out and instead of this chat message input let's use the content idea task we
21:20 created we'll replace this with trigger manually We'll add a call another
21:30 node choose get content ideas from the list there we go now let's add a split
21:34 out node drag in the suggestions from the idea step now let's add a loop to Loop over
21:42 each item awesome now for each of the suggestions that was generated we'll
21:49 trigger the AI agent flow but because the AI agent expects an AI chat input we
21:55 just need to add a temporary step here so once again edit fields
22:00 this time let's use Json and we want an object that looks like
22:07 this for the session ID we use an inbuilt variable called item
22:13 index the action should just be send message and the chat input will be the
22:17 suggestion we got from the idea generation step perfect we can feed this directly
22:23 into the AI agent and after each Lube is done we need to make sure to go all the way back
22:31 to continue with the next item which is idea now all we have to do is Click test
22:40 workflow and see the whole thing in action the AI agents will start
22:46 working and a bit of back and forth and finally off to feed five where
22:52 the post will appear under drafts highquality posts that are on
22:57 brand using the right topics and using the right style of riding fully
23:03 automated isn't that just awesome also in case you didn't know
23:04 FounderStack & workflow files
23:07 feed hiive which is the social media management tool I used here is one of
23:11 the sad products my team and I run and if you think all of this looked way too
23:15 overwhelming and complex feed hiive actually has a lightweight version of
23:19 this exact flow built-in you can write your posts ask feed's AI to give it a
23:26 grade and it will give you feedback and suggest changes to make your post better
23:31 to it's not using a fine-tuned model trained on your specific content so it's
23:35 not as good as a custom setup like the one we built here but it's pretty solid
23:40 and a very userfriendly experience you can get lifetime access to feed hiive as
23:45 well as three other tools my team and I run for a single onetime purchase they
23:50 all come with apis and automation they're all AI agent compatible go to
23:56 founder stack. proo to check it out I'll leave a link in the description now I
24:00 know I ran through some of this pretty fast but I have put the full workflows
24:04 together for you to easily import into your own NN account and a step-by-step
24:09 tutorial on how to set this up in an ocean dock if you are a Founder stack member
24:16 you can go and access this right away I have left the link in the description
24:19 below all you have to do is enter the email you used to buy founder stack and
24:24 you'll get access to this immediately if you want a more slow paced technical
24:29 introduction to n8n just Search Here on YouTube there are many excellent videos
24:32 from creators showing you how to build all kinds of cool automation with these
24:37 agents now I picked one example for this video which is content creation but
24:41 through this I'm using similar workflows for a whole range of different tasks I
24:45 hope you got inspired thanks for