The GPT-4o Viral Social Media Content System

Stamps ⌚️
00:00 Intro
00:40 System Overview
05:05 What you need to do this
10:51 Start Building
16:11 Perplexity
19:38 Summariser
21:42 Midjourney
23:48 Socials
26:19 Training on Data
28:58 Tone of Voice
31:17 Socials part 2
35:16 Airtable
44:18 Short Scripts
49:39 Run through

[Itssssss_Jack]

Would you like a FULL video on training a GPT on your stuff? 🔥

[StephanieIzzy]

I find the results for the agents are better when you toggle on "file search" in the settings of each place where you use the agent. This forces the agent to always use the files that are uploaded. Otherwise, it can decide to skip the file and just work with the instructions provided in the agent.

Also, I like the way this is setup, but as stated a few days ago I thought that the results were not really on point with my tone of voice, so I built an agent that is checking the results of the other agent and provides feedback on it based on the tone of voice rules. I divided it in: Objective: reached yes/no Check: what it needs to check for with this rule Verification: how is this rule implemented in this text. When the objective is not reached It will a suggested revision and a revision.

After that I use the same content agent that created the content for me to create a revised version based on the feedback provided.

Maybe this was already obvious to you, but otherwise hopefully this helps anyone.

[mike-leembruggen]

Amazing to see how GPT 4 can be used to create viral content across multiple platforms! Great video Jack!

[NightOwlMedia]

Another good solid Social Media automation. I can highly recommend your Skool community 👍

[AskTheRealDrBrown]

Great stuff. Joining the community this weekend

[magicanexus]

The value you bring here is awesome :O

[larsr.]

Very helpful video, but jumping over the Midjourney-bot makes it difficult to follow for first time viewers who did not see videos explaining this little but important detail.

[tgelenske]

Amazing content!! Is it possible to create the table into Google Sheets instead of Airtable? If not, why?

[DabeerArshad-vp]

Amazing video and I must say very well explained. However, I just want to know how did you create an the midjourney AI assistant. (I am new to AI, coding and automation), Could you please help me here. Or give me the video which had explained that part.

[ArtemSFG]

Hey Jack, thanks for sharing this super useful tutorial. Just one question: is there a reason we don't skip HTTP and Text Parser modules and input the article URL right to Perplexity? Perplexity can access the article. What are HTTP and Text Parser modules for?

[sr.modanez]

Thank you for your videos, they are very good, from a fan of yours in Brazil

[Emily-Bo]

Do you teach how to build an assistant in your community?

[aizingnomad]

Pretty awesome stuff 🤩 may i ask what's the average cost for running this 30 days. Make, Airtable and API cost? 🤔

[Drenchkat]

+1 subscriber. Thanks for the help brother!

[SabineKaluscha-og]

Very, very cool automation. I have noticed two problems.

The text that comes from RSS is sometimes too long and an error occurs. Can this be controlled?
When I set multiple entries to 'approve' in Airtable at the same time, they are all posted simultaneously. How can I set it so that all approved entries are posted sequentially, e.g., one post per platform per day?

[Opeyemi.sanusi]

Please a full video on tone of voice please 🙏🏾

[RJMCTV]

Really enjoyed this. New to Make. Not new to OpenAI or MJ. Question. Presume it’s the case that if I use the automation that creates the MJ prompt, then I can’t automate the posting using the APPROVED process, as I would separately need to create the MJ image using the prompt generated, and post that manually along with the Make created caption? Right?

[ericbreitkopf]

Could I have this system scrape for the content I want and create a video w an ai presenter and schedule to post on social media?

[StephanieIzzy]

I find that even if I give the tone of voice document in the agent the instructions to exclude certain language it still picks it up in the end result. I mean I excluded things like the typical language like maximizing, embark, embrace, demystified, unleash etc.

Do you have any ideas on optimizing this?

[TracyEvansProductions]

Thanks for another amazing workflow. It’s almost exactly what I need. But I need to use my own imagery. You mentioned having a workflow of that. Is there a video on it?