Leon, I can’t tell you just how much I’m learning watching all these videos. Your presentation style and clear instructions make it a breeze to replicate and customise for my own use cases. Thank you very much!
Love your presentation style Leon, simple and clear. Such a refreshing change from most of the content creators here. Also great content. I just started on Flowise and learning a ton from your videos. Thanks for all the work.
Welcome to the channel 👍. FW is definitely worth learning. I've been using it on actual client projects for nearly 2 years now and it's super reliable.
Thank you very much for the series of Flowise tutorials you provided, which helped me quickly build my own AI application from being a novice. Your explanations were very detailed and accurate. Thank you, and I look forward to you releasing more videos about Flowise in the future.
Another great presentation. I appreciate you showing the Google project API key setup and how to gain access to the respective "tools". I usually get a bit lost in there and your tutorials help a lot.
Your videos are awesome! Your snippets in the description are wrong though. after v3 you have ... and are missing part of the url such as "search?part=snippet&q="
Hi Leon, some of the new Flowise updates are nice. for example the powered by can be done straight in the embedding code which is awesome. Also GPT-4o mini is a game changer price wise as I still think out of all the models out there, openai still gives the best results and assistants are so easy and cool. With GPT-4o mini it is not only fast but considerably cheaper. I just wish the would get the reasoning steps with agents visible on the deployed versions. Keep up the great work.
Thanks Leon! The was a fantastic example for understanding the Supervisor/Worker relationship. I would have been helpful to add the detailed examples of the output in the comments, but that is minor. Thanks for your efforts.
This also rocks!! A little late to the agent party..im working my way through the series! Always refreshing to watch your videos …no click bait fluff here 😂😂 ps I think that rate limit is for trial keys?.
This is so great that I test everything on your videos in my flows. Can you make a video that will show the same workflow but in Langflow? There is no supervisor and worker and It is interesting how to build the same flow. Thx.
Hi Leon, another amazing video. Thank you so much for sharing your knowledge!!! I know I asked this question before and reached out to many people with this question: how can we build "human in the loop" functions into the workflow. I'm especially interested in figuring out how to ensure end users provide feedback on each component generated by each agent for improvement until satisfactory before the agent passes the task to the next agent. I would appreciate your insights very much!!!
Thank you for the feedback 💓. Human in the loop is not supported in Agentflows [yet]. It's definitely possible in Langgraph (the underlying technology behind agentflows) so I won't be surprised to see it added soon. I'll definitely create a video as soon as it's available.
I have tried to duplicate (more or less) your scenario but the file is not generated. I have deployed the flowise in 2 differents scenarios, one in Azure in the second in docker but my doubt is how I have to setup the path under the component for storing the file created in the tool Write File? Because I have written /home/user/flowise (in the path field) but althought the chat tells me that the file with a random name has been generated I can´t find it anywhere. What is wrong here?
Thanks Leon, You have done a fantastic job on creating all these videos. Unfortunately, I am having a problem with the google youtube API isn't working? When I try running the flow it comes back with null results. When I look at what results the api is getting in the youtube tags in the chat, all the OUTPUT is " " empty. I wonder if anyone else is having this problem. I have run the api from my terminal with the curl command and it works fine.
leon, youtube tools saying no valid json object found and when i clicked on searches it maid it resulted in like, no output is coming from the google cloud, what could be the error? do i need to buy any credit
Hi. Thank you for this great video. I followed many of your video a while ago and never had any issues. At this time I create simple credentials for openai in Flowise just using my API key. But now I added an organization in my openai account so I need an organization id parameter plus an API key to make any openai API call. But i can't find any way to add an organization id in the Flowise credentials. I checked in code without flowise, I get an error 429 without the organization id but it works perfectly if i add this parameter to the call. Any idea? naybe organization id are not supported by Flowise yet?
Leon help me please! Greeting messages are duplicated when I embed chat on my website. I've already changed the vector bank, prompt and everything else I could! Any idea what it is?
Amazing content Leon, thank you very much. I wonder if we can integrate LangGraph in Flowise. That would be great to ensure a more formal flow with loops and iterations, like a state machine.
@Leon van Zyl have you encountered message abruptly stopping when using the agent workflow? It seems to be an issue many people are having. Also can you create a video how to properly use the custom tool? I'm fetching data through the custom tool but it only shows up in the console. The agent can can't seem to retrieve the data.
Great explainer (Video) to the point I can’t tell you how many other RUclipsrs I’ve watched automation videos and it’s dreadfully long and you have to use five different platforms to get this task completed how do you feel about using perplexity? I feel that perplexity does a better job searching the web what’s your Outlook?
Great video Leon! Could you please make a video on connecting Flowise with WhatsApp? Also, why is Flowise/Vectorshift/Langflow your preferred platform instead of n8n for instance?
n8n is on my list of platforms to cover SOON 😂. I use it all the time. I'm exploring the features, strengths and weaknesses of as many platforms as possible.
Awesome thank you! Im looking but cant find it in your library but do you have a tutorial on how to upload a video file to one of these models using a tool like the "read file" tool? I'm wanting to implement Google Vertex AI Gemini 1.5 Pro's ability to watch video. Being able to upload files other than documents would be super useful for content creators.
@@leonvanzyl Looking at my VertexAI console now. I can attach video via direct upload (7MB limit), Google Cloud Storage (no size limit listed), and Google Drive (no size limit listed) - I wonder if its a matter of syntax for the correct url or cloud storage identifier in an input node?
Thanks for this tutorial! I'm trying to improve the tutorial by having the system email me the report instead of writing it to a file. Do you have a tutorial on an email custom tool?
@@leonvanzyl Thanks for this! Was there a reason you had Flowise call Make rather than write up an email send through NodeJS? Just wondering if it's not supported rather than Make was easier to do
👍🏻 do you have any videos about adding OpenAI assistants into a workflow like this? I’ve added a couple of mine in but can’t quite see how to connect them to each other. Can the supervisor be an OpenAI assistant or does it always need to be more generic?
Hey there! Sounds like an interesting use case. May I ask why you would prefer to use OpenAI Assistants API instead of just using these worker agents with RAG and tools?
@@Ethiventure Just copy the prompts you used for your assistants over and use the same model. You then have what would be the same those that you know.
I wish the docs were available at the time! 😅 I usually keep an eye on their Discord channel and X posts to find out when new features are released. I then dig through their code in GitHub to see the underlying Langchain/ LangGraph solution. Then it's trial and error until I'm ready to create a video on it 😬. These vids are effort
@@leonvanzyl It was working well for me, but now I get this error when calling the YT API: Request had invalid authentication credentials. Expected OAuth 2 access token, login cookie or other valid authentication credential. Making the call to the API endpoint on Postman gives the same error. Doing it on the browser works fine.
The huge issue I have found with this is that it only works inside Flowise. If you try it with a shared bot or embeded bot or API, it takes forever then outputs everything as 1 huge block of text. It does not output / stream in the same way as inside flowise test chat as you have here. Rendering useless for production. Ultra disappointing, unless Leon you know a way? Then that will be the most valuble video ever! Other than that this does not have much of a function other than personal use inside Flowise.
In my experience, the response does stream in the embed and API. I'll double check though. I agree that streaming is critical for this to be production ready.
![I.N "HALLUCINATION" | [Stray Kids : SKZ-PLAYER]](http://i.ytimg.com/vi/n5B5q1Hwt_U/mqdefault.jpg)
Leon, I can’t tell you just how much I’m learning watching all these videos. Your presentation style and clear instructions make it a breeze to replicate and customise for my own use cases. Thank you very much!
Thank you Matt! Just glad I could help 😃
I don't know, why you don't have 100 thousands of followers. Your Tubes are fantastic. Thx for sharing your knowledge Leon. I have learned so much !!!
I know, right ☺️.
Thanks for the kind words.
Love your presentation style Leon, simple and clear. Such a refreshing change from most of the content creators here. Also great content. I just started on Flowise and learning a ton from your videos. Thanks for all the work.
Welcome to the channel 👍.
FW is definitely worth learning. I've been using it on actual client projects for nearly 2 years now and it's super reliable.
Thank you very much for the series of Flowise tutorials you provided, which helped me quickly build my own AI application from being a novice. Your explanations were very detailed and accurate. Thank you, and I look forward to you releasing more videos about Flowise in the future.
Thank you for the feedback
wow. This is so good. Leon is practically devleoping the documentation for Flowise :)
I think so too! 😂
Another great video. Thank you Leon.
This is amazing!! Thank you for the support ❤️🙏
You're an absolute legend, Leon. Great content as always. Much appreciated. A++ 🙏
Thank you 😊
Leon, this is fantastic! 🤓 ✅ Flowise is quickly becoming my favorite go-to AI automation platform. Keep up the good work! 🎉
Thank you!
Another great presentation. I appreciate you showing the Google project API key setup and how to gain access to the respective "tools". I usually get a bit lost in there and your tutorials help a lot.
Thank you!
I'm really trying to simplify custom tools.
Outstanding video! You have an amazing talent for making things very clear to your viewers. One of the best I've ever seen. Thank you!
WOW, this is amazing feedback! THANK YOU!
Fantastic! I experiment extensively with Flowise but always learn new things from your videos.
Thank you. Keep them coming ...
Thanks Gilbert! Glad you veterans can still get some value from my videos 😀. Are you on the Flowise Discord?
@@leonvanzyl Yes. I am.
You are genius with Flowise. Thank you.
Thank you 😊. Just trying to give you value
Your videos are awesome! Your snippets in the description are wrong though. after v3 you have ... and are missing part of the url such as "search?part=snippet&q="
Finally I found this super coooool episode. amazing, thank you soooooo much Leon.
You're welcome 🤗
Very good. I have a different application for which I can use this structure. Keep making great videos, Wayne
Thanks, will do!
Hi Leon, some of the new Flowise updates are nice. for example the powered by can be done straight in the embedding code which is awesome. Also GPT-4o mini is a game changer price wise as I still think out of all the models out there, openai still gives the best results and assistants are so easy and cool. With GPT-4o mini it is not only fast but considerably cheaper. I just wish the would get the reasoning steps with agents visible on the deployed versions. Keep up the great work.
❤ this one! May be my favourite…outstanding content as always
Thank you!
Thanks Leon! The was a fantastic example for understanding the Supervisor/Worker relationship. I would have been helpful to add the detailed examples of the output in the comments, but that is minor. Thanks for your efforts.
Thank you for the feedback 🙏
Freakin’ brilliant, as usual! Leon for president!
Thank you 😂
You are one of the best i have come across when it comes to teaching flowise :) if you can cover 1 SQL tutorial using flowise that would be awesome..
Thank you!
Will do.
This also rocks!! A little late to the agent party..im working my way through the series! Always refreshing to watch your videos …no click bait fluff here 😂😂 ps I think that rate limit is for trial keys?.
You're not late. Now is the perfect time to learn about agents, and especially multi agent solutions 👍.
This just blew my mind!
It's awesome, and a lot of fun to play with.
This is so great that I test everything on your videos in my flows. Can you make a video that will show the same workflow but in Langflow? There is no supervisor and worker and It is interesting how to build the same flow. Thx.
WOW!!! This is absolutely awesome!!!
Great video Leon. I would like to see one multiagent video with VectorShift, if possible. Thanks
Hi Leon, another amazing video. Thank you so much for sharing your knowledge!!! I know I asked this question before and reached out to many people with this question: how can we build "human in the loop" functions into the workflow. I'm especially interested in figuring out how to ensure end users provide feedback on each component generated by each agent for improvement until satisfactory before the agent passes the task to the next agent. I would appreciate your insights very much!!!
Thank you for the feedback 💓.
Human in the loop is not supported in Agentflows [yet]. It's definitely possible in Langgraph (the underlying technology behind agentflows) so I won't be surprised to see it added soon.
I'll definitely create a video as soon as it's available.
@@leonvanzyl Thank you so much. I literally can't wait to see a new video. It's gonna be SO POWERFUL!!
Your videos are just too good. Thank you!
Thank you 🙏
I have tried to duplicate (more or less) your scenario but the file is not generated. I have deployed the flowise in 2 differents scenarios, one in Azure in the second in docker but my doubt is how I have to setup the path under the component for storing the file created in the tool Write File? Because I have written /home/user/flowise (in the path field) but althought the chat tells me that the file with a random name has been generated I can´t find it anywhere. What is wrong here?
Thanks a lot for sharing these videos, you are helping me a lot.
You're welcome!
Thanks Leon, You have done a fantastic job on creating all these videos. Unfortunately, I am having a problem with the google youtube API isn't working? When I try running the flow it comes back with null results. When I look at what results the api is getting in the youtube tags in the chat, all the OUTPUT is " " empty. I wonder if anyone else is having this problem. I have run the api from my terminal with the curl command and it works fine.
One question: The write to file tool only works if you are running Flowsie locally or does it work also if you deploy it to Render, Railway, etc?
leon, youtube tools saying no valid json object found and when i clicked on searches it maid it resulted in like, no output is coming from the google cloud, what could be the error? do i need to buy any credit
Hello Leon.
OK?
I need your help. I set up the flow, but when I ask to do the search it just calls the title creator and stops. How do I resolve this?
Hi. Thank you for this great video. I followed many of your video a while ago and never had any issues. At this time I create simple credentials for openai in Flowise just using my API key. But now I added an organization in my openai account so I need an organization id parameter plus an API key to make any openai API call. But i can't find any way to add an organization id in the Flowise credentials. I checked in code without flowise, I get an error 429 without the organization id but it works perfectly if i add this parameter to the call. Any idea? naybe organization id are not supported by Flowise yet?
Leon help me please! Greeting messages are duplicated when I embed chat on my website. I've already changed the vector bank, prompt and everything else I could! Any idea what it is?
Can we have a multi agent flow which can ask some information to user or have some human intervention like yes or no.
Amazing content Leon, thank you very much. I wonder if we can integrate LangGraph in Flowise. That would be great to ensure a more formal flow with loops and iterations, like a state machine.
Thank you!
Agentflows are actually using LangGraph under the hood. I know the Flowise team is going add more functionality to this soon.
Leon, it's really really helpfull! ❤
You're welcome 🤗
@Leon van Zyl have you encountered message abruptly stopping when using the agent workflow? It seems to be an issue many people are having. Also can you create a video how to properly use the custom tool? I'm fetching data through the custom tool but it only shows up in the console. The agent can can't seem to retrieve the data.
Great content as always. Do you have or can you make a video on how to do this with VectorShift?
You're welcome 🤗.
Will see what I can do 👍
Great explainer (Video) to the point I can’t tell you how many other RUclipsrs I’ve watched automation videos and it’s dreadfully long and you have to use five different platforms to get this task completed how do you feel about using perplexity? I feel that perplexity does a better job searching the web what’s your Outlook?
Thank you for the amazing feedback! I try my best to improve with each video.
Perplexity is awesome. Maybe I should create a video of some sort on it.
@@leonvanzyl can make one showing how to answer emails in outlook/gmail.
GREAT VIDEO!!!
Great video Leon! Could you please make a video on connecting Flowise with WhatsApp? Also, why is Flowise/Vectorshift/Langflow your preferred platform instead of n8n for instance?
n8n is on my list of platforms to cover SOON 😂. I use it all the time.
I'm exploring the features, strengths and weaknesses of as many platforms as possible.
@@leonvanzyl thanks!
Brilliant! However, after updating the Supervisor prompt and running the flow, the screen went white. Before the change, it was running like a champ.
Wow, haven't seen that myself. Are you running locally or cloud?
@@leonvanzyl The problem occurred locally, I just moved it to render, and it's working as expected.
Awesome thank you! Im looking but cant find it in your library but do you have a tutorial on how to upload a video file to one of these models using a tool like the "read file" tool? I'm wanting to implement Google Vertex AI Gemini 1.5 Pro's ability to watch video. Being able to upload files other than documents would be super useful for content creators.
I don't think video files are supporting yet..
Hmm.. I better confirm this. Would be awesome!
@@leonvanzyl Looking at my VertexAI console now. I can attach video via direct upload (7MB limit), Google Cloud Storage (no size limit listed), and Google Drive (no size limit listed) - I wonder if its a matter of syntax for the correct url or cloud storage identifier in an input node?
Thanks for this tutorial! I'm trying to improve the tutorial by having the system email me the report instead of writing it to a file. Do you have a tutorial on an email custom tool?
Check out this video where we use a tool to send emails.
ruclips.net/video/MaqcO15y-Vs/видео.html
@@leonvanzyl Thanks for this! Was there a reason you had Flowise call Make rather than write up an email send through NodeJS? Just wondering if it's not supported rather than Make was easier to do
@@villanianalytics you can definitely just write some node code to make it work. I actually did exactly that in an older video.
👍🏻 do you have any videos about adding OpenAI assistants into a workflow like this? I’ve added a couple of mine in but can’t quite see how to connect them to each other. Can the supervisor be an OpenAI assistant or does it always need to be more generic?
Hey there!
Sounds like an interesting use case. May I ask why you would prefer to use OpenAI Assistants API instead of just using these worker agents with RAG and tools?
@@leonvanzyl because I already have them and feel like I ‘know’ them and it seems like it should be a bit simpler than doing the RAG process.
@@Ethiventure Just copy the prompts you used for your assistants over and use the same model. You then have what would be the same those that you know.
leon how u do ur research of all this? like is there any doc or something??
I wish the docs were available at the time! 😅
I usually keep an eye on their Discord channel and X posts to find out when new features are released.
I then dig through their code in GitHub to see the underlying Langchain/ LangGraph solution.
Then it's trial and error until I'm ready to create a video on it 😬.
These vids are effort
@leonvanzyl my god, thank you so much for all those efforts man, really appreciate ur work.
Can anyone give an example of how to use multi query retriever?
how much does it cost to run this team once?(token costs,api calls etc..)
It was about 7 to 11 cents (USD) with GPT-4o.
@@leonvanzyl Does it need GPT-4o for the workers. Would swapping out for gpt-3.5-turbo-0125 reduce the costs significantly ?
Cool! you are great.
I love you Leon.
Why didn't my bot respond after I followed your instructions? How can I troubleshoot this issue?
Thank you very much for answer my question!
Did you notice any errors in the terminal?
Very nice!
I have this message at the end when enter the keywords and run " Message stopped", How can I managed it?
I noticed that message when my OpenAI credits were depleted. Check your OpenAI account to see if that's also the case for you.
@@leonvanzyl Instead I have enough OpenAI credits, I don't really understand what is going on
Leon jy is bakgat ✋🎤 awesome
Dankie Manie!
very nice！and i hope a vedio which could teach me build a data analysis team with flowise
Check out my other videos on Agentflows. The last video includes data analysis from a PDF document.
@Leon ive sent you several emails about hiring you to help me with a project, to no avail. are you not for hire?
Hey there, did you send the mails to Cignaitiv? I'm admittedly behind on responding to maiils, so really sorry about that. Will get you yours soon 😁
is it possible to use ollama 3.1 8b ?
I recommend going with the 70b model instead. The 8b model is not great for agents.
You can use Groq to run the 70b model for free.
Thanks for your amazing job ! I wanted to join your discord but it seems to be full of X content spams.
Yeah, I seriously need mods.
Most likely going to can the Discord channel 😬
Hi! anyone knows how to solve this>>>Cannot read properties of undefined (reading 'length') let me know
Did you remember to save the flow before running it? 😁
I usually see that message if I don't save
@@leonvanzyl yes ... I saved but is not working
Amazing
its not using yt api cant understand why
Is it calling the custom tool at all? If not, then instruct the worker in the prompt to use the custom tool.
@@leonvanzyl It was working well for me, but now I get this error when calling the YT API: Request had invalid authentication credentials. Expected OAuth 2 access token, login cookie or other valid authentication credential.
Making the call to the API endpoint on Postman gives the same error. Doing it on the browser works fine.
👏🏼
👍👍👍
i need leon´s brain
😂
nice work!! i love this.
The huge issue I have found with this is that it only works inside Flowise. If you try it with a shared bot or embeded bot or API, it takes forever then outputs everything as 1 huge block of text. It does not output / stream in the same way as inside flowise test chat as you have here. Rendering useless for production. Ultra disappointing, unless Leon you know a way? Then that will be the most valuble video ever! Other than that this does not have much of a function other than personal use inside Flowise.
In my experience, the response does stream in the embed and API.
I'll double check though.
I agree that streaming is critical for this to be production ready.
Hi @@leonvanzylDid you find anything, would like to see your results. Thanks for your videos ...