Errata: At 32:12, I do an overview of the front-end code. There, you see some code that we fixed two minutes before, when I said "hello from the future" (the initialization of our Crew object). Don't pay too much attention to that. Anyways, you can ask me your questions here! I hope you find this useful!
Oh yea! It's like Bob Ross for coding AI. Great to have another video that can be expanded and reused for other applications. The emphasis to always have a front end UI is what sets these videos apart even if it takes a little more time. Bravo as usual.
hey Brandon! i appreciate it, thank you! i'm glad you are finding this series useful :) please let me know if you have any suggestions or feedback. all the best!!
@@alejandro_ao Quick question, when setting a task with parameter of 'human_input=True', is there a way to enable this human interactive in the web frontend? Thanks in advance.
Great complete example! Thanks for sharing. Wouldn’t it be even better to have the UI code provide the callback to the call so the crew itself is oblivious of rendering?
First of all, I really congrats you bro, your ideas are very useful thanks again. However, I am very curious about your opinion on the following as a custom agent tool structure: “For example, how can you plan a code evaluation agent structure?” For example, .java code will be entered as input, the execution of the code will be evaluated with a certain weight, the proper writing of the code will be added as an evaluation parameter, and the use of commands will also be added as an evaluation parameter, and ultimately a custom agent that returns a score. How do you think we can do it? You usually make web search tool examples, for example, wouldn't this be different? Wouldn't it be great if you prepared such a content video? I'm curious about your ideas on this subject! For example, the first stage can be considered as a simple two-stage evaluation: 1- The presence or absence of a syntax error in the code (Output: True or False, accordingly 1 point if true and 0 if false) 2- Is there a Comment in the code or not? (like condition 1, if the output is true, 1 point if not 0) How about an agent and a custom tool that will evaluate only according to these two structures?
At 26:42, the st.components.v1.html section was commented out. I may have missed where you went back to fix it. Do you have a time stamp I can look at?
hey there, good point! sorry about that, the component was not working when i was recording so i just left it commented out. it was only just a quick preview of the newsletter. however, support for that streamlit component is dropping. i believe that is why it did not work. if you want to show html code you can use `st.html`.
hey there! thanks man. i would very much like to do a video about that indeed. thanks for the recommendation! i will include it in a future series, definitely coming up in the next few weeks. thanks!
This is great! How would u turn this into a backend api? Since this is just a streamlit application, ideally I want my crew AI in a fast api framework that I can use different front ends for. How would I do that?
hey there, i have a pretty rudimentary tutorial here: ua-cam.com/video/74c3KaAXPvk/v-deo.html in it, i show how to deploy a streamlit app like the one you see here
Hi Alejandro, a great tutorial! Did you try to combine crewai, Exa search engine, as a client, with Azure OpenAI? How I could do that correctly? Thanks!
yes totally! i haven’t used crewai with azure openai so but it should work just as well as with the openai api :) it would work very similar to this other video when i use exa and crewai for a research agent crew 💪 just define your exa tools, and specify the llm client that runs behind leach agent!
@@alejandro_ao Hi Alejandro, do you have this demo/tutorial version that does not use poetry? I think it is an extra dependence that could be, in principle, avoided. Thanks!
@@RicardoRocha00 oh right! yeah crewai has evolved a lot since this video. i really need to make an updated one. thanks for bringing my attention to this! you can expect that very soon!
hey rodrigo! i’m going to be publishing the other part of this tutorial later this week (which contains how to create that crew.py file). i’m working on finishing off that part with the people from Exa to make it as faithful as possible to their service. i can ping you once that part is up!
you will need a LLM to do this. and for that, you most likely need an API key of your LLM provide, unless you are running this locally. however, if your concern is the price, you can use GroqCloud as your LLM provider. They offer a free tier that should be more tangible enough for educational purposes. Switching to this one instead of OpenaAI’s models should involve changing a couple of lines in your code: python.langchain.com/v0.2/docs/integrations/providers/groq/
oh my bad!! i will update this with the full GH repo once I finish next video (i am working on it with the people from Exa). For now, here is the code that we covered in this video 👇 - Front-end of the app: gist.github.com/alejandro-ao/9f917ed27a55aad5aef277fb51993449 - Step-callback function: gist.github.com/alejandro-ao/6b5cd6166f6d9219c26222809bcd8392
Errata:
At 32:12, I do an overview of the front-end code. There, you see some code that we fixed two minutes before, when I said "hello from the future" (the initialization of our Crew object). Don't pay too much attention to that. Anyways, you can ask me your questions here!
I hope you find this useful!
Brother you are a legend 🙌🏾 I've been needing a UI for crews. You still top of the game! Thank you for taking the time to make and share this
thanks Andy! i’m glad to be able to contribute to your learning path, regardless of how small my contribution is :)
Oh yea! It's like Bob Ross for coding AI. Great to have another video that can be expanded and reused for other applications. The emphasis to always have a front end UI is what sets these videos apart even if it takes a little more time. Bravo as usual.
hey Brandon! i appreciate it, thank you! i'm glad you are finding this series useful :) please let me know if you have any suggestions or feedback. all the best!!
Great to see another quality video from you.
glad to see you here!
Thanks you so much!
you are amazing, thank you so much Ray
Thanks!
You bet! thanks so much for the tip!
This is such a great tutorial journey.
thanks for being part of this journey!
@@alejandro_ao Quick question, when setting a task with parameter of 'human_input=True', is there a way to enable this human interactive in the web frontend? Thanks in advance.
Great complete example! Thanks for sharing. Wouldn’t it be even better to have the UI code provide the callback to the call so the crew itself is oblivious of rendering?
been waiting for new vidoes this is lit! 🔥🔥 Thanks for vid.
🫶
First of all, I really congrats you bro, your ideas are very useful thanks again. However, I am very curious about your opinion on the following as a custom agent tool structure: “For example, how can you plan a code evaluation agent structure?” For example, .java code will be entered as input, the execution of the code will be evaluated with a certain weight, the proper writing of the code will be added as an evaluation parameter, and the use of commands will also be added as an evaluation parameter, and ultimately a custom agent that returns a score. How do you think we can do it? You usually make web search tool examples, for example, wouldn't this be different? Wouldn't it be great if you prepared such a content video? I'm curious about your ideas on this subject!
For example, the first stage can be considered as a simple two-stage evaluation:
1- The presence or absence of a syntax error in the code (Output: True or False, accordingly 1 point if true and 0 if false)
2- Is there a Comment in the code or not? (like condition 1, if the output is true, 1 point if not 0)
How about an agent and a custom tool that will evaluate only according to these two structures?
At 26:42, the st.components.v1.html section was commented out. I may have missed where you went back to fix it. Do you have a time stamp I can look at?
hey there, good point! sorry about that, the component was not working when i was recording so i just left it commented out.
it was only just a quick preview of the newsletter. however, support for that streamlit component is dropping. i believe that is why it did not work. if you want to show html code you can use `st.html`.
Another quality video from you! 🔥 Could you also make a video on RAPTOR as it outperforms RAG on longer documents?
hey there! thanks man. i would very much like to do a video about that indeed. thanks for the recommendation! i will include it in a future series, definitely coming up in the next few weeks. thanks!
@@alejandro_ao thanks man. I know just like your other content, it will be a real value-add. Looking forward to it.
This is great! How would u turn this into a backend api? Since this is just a streamlit application, ideally I want my crew AI in a fast api framework that I can use different front ends for. How would I do that?
i am working on a video about this! :)
@@alejandro_ao GREAT! I’m hype for it. Do you think it’ll release this week?
I am waiting for your video that could automate statistical data analysis.
cool video! Will you plan for a video on deploying LLM applications
hey there, i have a pretty rudimentary tutorial here: ua-cam.com/video/74c3KaAXPvk/v-deo.html
in it, i show how to deploy a streamlit app like the one you see here
Hi Alejandro, a great tutorial! Did you try to combine crewai, Exa search engine, as a client, with Azure OpenAI? How I could do that correctly? Thanks!
yes totally! i haven’t used crewai with azure openai so but it should work just as well as with the openai api :)
it would work very similar to this other video when i use exa and crewai for a research agent crew 💪
just define your exa tools, and specify the llm client that runs behind leach agent!
@@alejandro_ao Hi Alejandro, do you have this demo/tutorial version that does not use poetry? I think it is an extra dependence that could be, in principle, avoided. Thanks!
Thank you for this video 🙏😊!
Hey Adelina! It's my pleasure 😊
i think i followed your steps but and the chat on the GUI is not appearing in my case.it seems the step_callback is not being executed. Any ideas?
hey there did you solve this?
@@alejandro_ao no. It seems there were changes with the call back functionality. It's very frustrating!
@@RicardoRocha00 oh right! yeah crewai has evolved a lot since this video. i really need to make an updated one. thanks for bringing my attention to this! you can expect that very soon!
Hey Alejandro. Sorry if this a simple question. I'm a initiating at all of this.
Did you shared the crew.py file? Didn't find at you GH.
hey rodrigo! i’m going to be publishing the other part of this tutorial later this week (which contains how to create that crew.py file). i’m working on finishing off that part with the people from Exa to make it as faithful as possible to their service. i can ping you once that part is up!
@@alejandro_ao awesome! Thanks. Your content is Amazing and it’s helping a lot here
very good
it is possible without api key or another approach??
you will need a LLM to do this. and for that, you most likely need an API key of your LLM provide, unless you are running this locally. however, if your concern is the price, you can use GroqCloud as your LLM provider. They offer a free tier that should be more tangible enough for educational purposes. Switching to this one instead of OpenaAI’s models should involve changing a couple of lines in your code: python.langchain.com/v0.2/docs/integrations/providers/groq/
This camera angle works better for you.
😎
top top top
Great!!!!
you are
will we get the code next week?
oh my bad!!
i will update this with the full GH repo once I finish next video (i am working on it with the people from Exa). For now, here is the code that we covered in this video 👇
- Front-end of the app: gist.github.com/alejandro-ao/9f917ed27a55aad5aef277fb51993449
- Step-callback function: gist.github.com/alejandro-ao/6b5cd6166f6d9219c26222809bcd8392
Thank you so much!
thank YOU for watching!