Hi All, Have been trying to get Brandon's CrewAI Rag code running that scrapes youtube for the five top videos (crewai-rag-deep-dive/2_youtube_and_web). I have installed and re-installed the poetry shell. The code is exactly the same - just a git clone of Brandon's code. That being said I have followed the vid twice to ensure all is well - I cannot see where my env starts to go pear shaped. My output seems different - less verbose? It is when the code starts executing the second agent/task that it fails to add any of the five videos to the vector store. I have let the code run to the end and it finishes all the tasks and does find a few items from the web searches. But it is the failed vectorDB that has me stumped! Help please! Thanks Alan. *** terminal output *** crewai-rag-basics-py3.10(base) a_shipley@ShipleyAI:~/Python_poetry/crewai-rag-deep-dive/2_youtube_and_web$ python crew.py Please enter the YouTube handle to analyze: bhancock_ai > Entering new CrewAgentExecutor chain... To proceed with scraping the latest five videos from the specified YouTube channel, I will use the available tool to fetch the latest videos from the channel handle `bhancock_ai`. Action: Fetch Latest Videos for Channel Action Input: {"youtube_channel_handle": "@bhancock_ai", "max_results": 5} videos=[VideoInfo(video_id='8PtGcNE01yo', title='CrewAI Flows Crash Course', publish_date=datetime.datetime(2024, 10, 25, 13, 45, 29, tzinfo=datetime.timezone.utc), video_url='https://www.youtube.com/watch?v=8PtGcNE01yo'), VideoInfo(video_id='cyPZsbO5i5U', title='Supabase Postgres Vector DB Crash Course', publish_date=datetime.datetime(2024, 10, 18, 20, 48, 14, tzinfo=datetime.timezone.utc), video_url='https://www.youtube.com/watch?v=cyPZsbO5i5U'), VideoInfo(video_id='2qU3SPPojDA', title='How to Build a NextJS MVP using v0, Claude, and Cursor', publish_date=datetime.datetime(2024, 10, 10, 17, 13, 35, tzinfo=datetime.timezone.utc), video_url='https://www.youtube.com/watch?v=2qU3SPPojDA'), VideoInfo(video_id='np1cwfDjCF4', title='Build a Full Stack AI Marketing Platform SaaS [Premium Course Announcement]', publish_date=datetime.datetime(2024, 9, 19, 17, 13, 33, tzinfo=datetime.timezone.utc), video_url='https://www.youtube.com/watch?v=np1cwfDjCF4'), VideoInfo(video_id='fMa2zQIkQwM', title='Claude 3.5 Crash Course for Developers: Code 10x Faster in 2024 [Claude 3.5 artifacts]', publish_date=datetime.datetime(2024, 8, 21, 13, 39, 24, tzinfo=datetime.timezone.utc), video_url='https://www.youtube.com/watch?v=fMa2zQIkQwM')]

Hey Ollama users. I have a 4200 word, Word document. It has some tables. I want to try some embedding with it. What Opensource models do you use for embeddings?

Hi guys. I'm a medium developer, but I can mostly get things to work the way I need in the end. However, the bane of my existence is the "production" side of things. Installing dependencies and all the stuff that makes my mind numb with dullness. Which begs the question: Does any one have a good guide or suggestion for deploying a crew to some place where it can be reached as an API (and no, CrewAI's on setup did nothing for me). 🙏

I often find that if you don't have a plan, you seem to lack direction. This can then lead to the failure of a product or process. I am a great believer in plans. Even if the plan does not achieve everything it set out to do, it steers you in the correct direction. As long as you realise from the outset that a plan must be flexible, you can implement corrective measures to mitigate potential risks. I also find that goal setting is essential. Always have a goal in mind. Not big goals. Go for the 'elephant analogy'. How do you eat an elephant? One bite at a time. This is easy to implement in software/product development as these processes are very iterative. I will plan out the elephant in something like excalidraw/miro etc. This is the big goal, the end product or as @Brandon Hancock calls it 'The North Star'. You then take this elephant and break it down into sections: A trunk, a leg, etc. These could be your Goals. You then break down the sections into tasks, which are lots of small, single-process tasks. Create a login button, attach login logic to the login button, add a login button to the interface, and test the interface. Task complete, strike it down, pick up the next and continue till the proverbial elephant has been consumed. When you have hit a major milestone in your project reward yourself with something. Maybe a book, maybe some family time, or a meal without the need for disappearing into the office (Sorry, wife). When you conclude your project deliverable, I will reward myself. It can be as large or small as you like. Don't use the revenue generated from your endeavours to ultimately reward yourself; re-invest in yourself and/or your company. I recently completed a consulting gig. It paid well, and all milestones and goals were achieved to the client's satisfaction. To my satisfaction, the money rolled in, and I 'treated' myself to a new 'toy' today. I now have the Operating system Holy Grail, Windows, Linux and Mac all available to test and play with

Hello, I saw a notice in the course saying that we should use nextJs 14 instead of 15. Can anyone explain why? I just installed nextJs and it assumed I wanted the most recent version. I'm happy to work through the course with version 15, but if it's going to cause me a ton of issues, I'd rather take the time now to go back to 14.