Thanks! I've had a lot of success with RSS.app for a similar use case of lots of articles across lots of sites. You can set up a system to import into Airtable and RSS.app has an easy system to build RSS feeds on websites that don't have them natively

We're getting HTML from Claude within this automation so it would be possible to send that to practically any CMS and it should look good. Webflow is supported within Make.com so it should be fairly straightforward. You'd just need to swap out the modules. Image uploads can vary a bit from system to system but might be easy enough to get working.

Ha! Thanks!

Thanks Stephen, appreciate it! Love your work

Ha! Not sure how I overlooked that module! That would have made life a bit easier alright. I think the benefit of going the "add slide" approach is if there's an unknown number of slides to add in a section you could put it in an iterator and add them dynamically. Like if you were asking GPT4o to output an array of slides for insights and there could be as many slides as the LLM deems necessary. That would be harder to do with a super fixed template ... Would probably need to delete unneeded slides. But yeah, the "Create a presentation" module is probably a better approach to cut down on the number of modules and operations. Thanks for that!

Hi, it's perfect for complete beginners and we've got plenty of people in our community that started in the same position as you with the same goals. We've got a full beginner section built out along with lots of automation templates in the community to work through. We're still building out our intermediate and advanced sections within the make course itself, but almost all of the concepts that we're planning to cover there are demonstrated in our automation template blueprint videos. You can get plenty of support direct from us in our live workshops or by asking questions within the community. Hope this helps! Alan

Hi, it's perfect for complete beginners and we've got plenty of people in our community that started in the same position as you with the same goals. We've got a full beginner section built out along with lots of automation templates in the community to work through. We're still building out our intermediate and advanced sections within the make course itself, but almost all of the concepts that we're planning to cover there are demonstrated in our automation template blueprint videos. You can get plenty of support direct from us in our live workshops or by asking questions within the community. Hope this helps! Alan

Thanks for that ... Yes, that solution would have worked here. I've created other Make automations where the structure of the input JSON is dynamic so the Create JSON module doesn't really work as a solution. That's where this hack came from and it's now my go to! Make really needs to come up with a "MAKE_JSON_SAFE" function. It's ridiculous that you need to formally create a "data structure" to simply escape some characters to pass into a web service call!

@@TheAIAutomators Yeah fully agree! When its dynamic you have a problem. This seems like an obvious standard tool that make should provide :) Love how you have a solution for dynamic JSONs as well. Btw also learning lots from you guys! You have great content. Keep up the good work.

Good question - The costs (static and variable) varies based on a number of factors. Make - $7 per 10,000 operations on Make - With 90+ modules combined with the webhook architecture, this scenario can be heavy on operations. Image Generation: Flux 1.1 Pro Ultra Costs $0.06 per image on Replicate.com Image-to-Video Generation LumaLabs costs $0.40 for a 5 second 24fps video at 720p Runway ML costs $0.25 for a 5 second 24fps video at 720p Creatomate The Monthly Plan of $49 / month is their cheapest plan This gets you 2,000 credits equates to around 2.5 hours of video rendering at 720p This is similar pricing to other solutions in the market like Json2video and Shotstack. (Creatomate doesn't have a pay as you go plan though) Music Generation: Suno via GoAPI On the Free GoAPI Plan it costs $0.04 per music track. However last weekend, GoAPI stopped supporting Suno as have hit issues with watermarking of content. They're working on a solution

Thanks! Yes, exactly, the url is just a unique key. As you said, it doesn't provide any additional meaning when it comes to retrieval. So the Airtable Record ID would work just as well. We have onboarding calls with all new members so happy to chat through it if you join :)

Thanks! We have onboarding calls with new members in the community, so would be happy to jump on a call to chat through the system :)

Ha!! Thanks so much!

Yeah it's a nice system! It's way more powerful than services like 0codekit

Glad you liked it! Yes, I'm ditching my own Ahrefs account here shortly as they've finally pushed me on to their newer pricing plan!

Thank you! Yeah I've played around with n8n a bit and it's great tool (especially as you can self-host it, so it's great if you're really working at scale, while also being great for bigger clients too). We'll may start using it at some point in our videos. A few people in our community use it heavily too. What do you like most about it? Alan

yeah for sure - we also have a video on how to instantly trigger make from Google Sheets, which you could also use in a Google sheets version of this automation to make it more streamlined: kzbin.info/www/bejne/d4WqqnVsqL2goKs - You may need to tweak the automation for image storage. You could potentially use Google Drive to store the images but I'm not 100% sure if you can upload these to Pinterest. If you can't pass Google Drive URLs to Pinterest then download the image within your make scenario from Google Drive and then select "Base64 image" instead of "image URL" when creating the Pin and then pass the data to that.

Hey, our social media AI automation is linked in the description. You'll understand the flow after watching that video. It uses DALLE-3 for images but I've adapted it slightly in this video to use FLUX.1 instead.

This varies based on your theme and any CSS styling. You could specifically add a custom CSS class to the img tag within the automation (eg. <img src="___" class="your_css_classname"...). Then you could add custom CSS to your theme to target that specific CSS class and change how it's presented like setting a max-width or adding whatever styling you want, or you can add some broader CSS styles on your website to change how all images are presented within the main content of your posts. ChatGPT is usually pretty good at creating CSS. Hope this helps! Alan

Good suggestion! We hope to cover that at some point. We've got two other videos on our channel that cover some of the concepts needed for this (an auto updating AI agent using OpenAI assistants API and the other using a more complex custom RAG approach)

Thanks! Yes Shopify allows creation of articles via their API and there's a Make app for it. It should also be able to grab an XML sitemap from the Shopify site for internal linking. I'm not 100% sure about what type of changes you'd need to make to the automation to get the images and KZbin embeds integrated like I'm doing in the video but it might be straightforward.

Ha! Thanks - Make seriously need to build a native function for this

@@TheAIAutomators Yes totally 👍Your code fixed a json bug I had with the Gemini API too 🙏 One question: What if the website also had 500 blog articles, would this data be upserted into the same Pinecone Db as the product info? E.g. Would all data to do with website X ( product page data, articles, company pages, how to guides etc) all go into one central Pinecone Db so it could then be efficiently queried by the AI model?

Yeah that's a good question. It really depends on your use case so you'd need to do some testing and trial and error with it. Having everything in a single vector store could result in no products being returned, for queries where you may only want products for example. So it may be better to have the different entities in separate namespaces and have logic around which vector store to call and when. (Similar to how AI Agents work with function calling) - There's no one-size-fits-all; it's specific to requirements really

Great question! Running everything locally or in a private cloud would be one way to go about it. FAISS is a local vector store that you could use instead of Pinecone. You can run local embeddings models and LLMs on Olama or LM Studio for example. Possibly tie it all together with an N8N instance locally. It's a more hands-on setup then what I showed in the video but a lot of larger companies do it. Another approach is the anomynisation of personally identifiable information before creating the embeddings or sending to the LLM. Then you may be able to use the likes of OpenAI on Azure cloud which supports stricter data security requirements.

You're very welcome!

Hey David. Great question. In case you haven't seen it, I did a video on using Pinecone and I talk about using OpenAI Assistants vs Pinecone at around the 5 minute mark - kzbin.info/www/bejne/en-nemeZaN9gbKM That breaks down the pros and cons of each. Hope that helps

You're very welcome and we'd love to welcome you to the community! On the re-ranking question, it really depends on your use case. There are dedicated lightweight reranker models both in Pinecone and on huggingface if your use case requires very high volumes of queries. These can be fine tuned if it's a specialized subject matter. If your use case is lower volume (like in this video), you could use an LLM to carry our the reranking. I've done a version of this in one of our micro templates in the community where I build an ecommerce chatbot on botpress. You can see it around the 36minute mark here - kzbin.info/www/bejne/ZmOqqYCuYruhirMsi=7-EWTjCutPYQD02e. In it I ask the LLM to review the 15 items returned from Pinecone and filter out irrelevant results (so it's more of a filter than a reranker) I also do a version of it in our news automated system where again, I ask it to provide a score based on an input and a context and if that score is over a certain threshold, the automation progresses. You can achieve a lot via prompt engineering and trial and error here - either in a single shot or multi-stage pipeline of prompts. If you were getting back 15 results for example, you could feed those in altogether and get the LLM to use chain of thought based on a detailed brief, to provide the rationale for the relevance of each item and then provide a score and output an array of objects in JSON. Then you can reorder based on those scores. Or if there's too much data, you can send them in smaller batches or one at a time and aggregate and rerank after. With this approach you could use any LLM really. LLMs are more sophisticated than Vector search which is usually done using the likes of cosine similarity and distances between vectors. So with the right prompting and a bit of patience, you should get some quality outputs. The LLM approach to reranking can slow down the output, which was a problem in the ecommerce chatbot example above. So speed and cost are considerations as well. On the question about local vector solutions vs Pinecone and if they're any good in terms of accuracy of results. In theory, if they use the same algorithms like cosine similarity, then they should be exactly the same. The variable is what embeddings model you use. Hope some of that helps Daniel