Praison AI Custom Tools: Integrate ANY Application with AI Agents
HTML-код
- Опубликовано: 21 май 2024
- 🌟 Welcome to an in-depth exploration of Praison AI Custom Tools! In today’s tutorial, we'll dive into the world of AI customisation and efficiency. 🚀
👉 What You'll Learn:
Introduction to Custom Tools: Understand the essentials of custom tools and their significance in enhancing AI agent performance.
Easy Setup Process: Step-by-step instructions on setting up your environment using YAML configuration and the init function.
Real-Time Demonstrations: Watch as we integrate internet search capabilities with our AI agents using simple coding techniques.
Practical Applications: Discover how these tools can be tailored to meet specific needs and significantly improve the output of your AI systems.
🔗 Timestamps:
0:00 - Why Use Custom Tools?
0:50 - Setting Up Praison AI
2:17 - Building Your First Custom Tool
3:55 - Integrating Tools with AI Agents
6:03 - Running the Complete System
🛠️ Get Started: Dive into the simplicity of low-code solutions and see how custom tools can revolutionize your AI projects!
👍 Like, Share, and Subscribe: If you find this video helpful, please like, share, and subscribe for more insightful content on Artificial Intelligence. Hit the bell icon to stay updated!
🔗 Resources:
Sponsor a Video: mer.vin/contact/
Do a Demo of Your Product: mer.vin/contact/
Patreon: / mervinpraison
Ko-fi: ko-fi.com/mervinpraison
Discord: / discord
Twitter / X : / mervinpraison
Code: github.com/MervinPraison/Prai...
#PraisonAI #Custom #Tools #FunctionCalling #LowCode
#PraisonAI #MultiAgentFramework #LowCode #LowCodeMultiAgentFramework #CreateMultipleAgents #CreateMultipleAIAgents #EasiestWayToCreateMultiAgents #AIAgents #AIAgent #MultiAIAgent #MultipleAIAgents #PraisonAITools #PraisonAITool #PraisonAIFunctionCalling #PraisonAIFunctionCall #PraisonaiTools #PraisonaiTool #PraisonaiFunctionCalling #PraisonaiFunctionCall #CrewAITools #CrewAITool #CrewAITools #AutogenTools #AutogenFunctionCalling 
Хобби
Nice to see you expanding your own tools, keep them coming
Thank you
Love your videos so much! Can you make a tutorial video on FlexClip’s AI tools? Really looking forward to that!
how to use llama3 7b locally with search engine?
Another great clarifying video, concepts and specifics, great combo, thank you, WR
Thank you
This is amazing!
Mervin,
I am a recent subscriber to your channel and I must commend your ability to distill complex concepts into easily understandable content. Your approach has been immensely beneficial to me as a newcomer in this field.
I am particularly interested in understanding how to handle various data formats such as PDF, CSV, and database files. Specifically, I am looking for an efficient way to upload these files (potentially in batches), perform necessary analyses, and subsequently purge all data post-analysis.
The rationale behind this approach is to ensure the utmost privacy and security when dealing with sensitive data. By limiting the data’s lifespan to a single session, I believe we can significantly mitigate the risks associated with long-term storage.
I would greatly appreciate any insights or suggestions you might have regarding this approach. Your expertise in this matter would be invaluable.
Thank you for your time and consideration. D.R.
Mervin, you made a MetaFramework, a new kind of programming ! very very smart, as all your videos... 2 questions :
- Why let us choose betwenn crewai and autogen ? Do you plan to add workflow features choices ? (erf.. 3 questions..)
- Would you one day show us an efficient RAG and/or long memory tool for Praison AI ? I guess we are plenty to look for that
🙏thx again and again for all your precious video !
1. Will be adding workflows so multiple framework can be used in one workflow . Each framework is good in certain things. So it’s better to run and choose your best
2. Sure in the near future will work on adding memory
Also will show efficient RAG and how to evaluate RAG in a dedicated video
Whaouuu.. As you always say, I'm so exiting about that! It's so amazing that you share all that with the world.. 🙏
That’s a great starting point for me. I am trying to create agentic workflow that would do internet research on specified websites and then a SEO agent would create summaries of the best articles for a certain timeframe (for example May 2024) and create titles and hashtags for these summaries with links to the original articles and another agent would be able to create / generate images from these summaries and the last agent would upload the article summaries with all contents and images to a blog on my Wordpress website. You think this could work? 🤔
Dude this is amazing!
Thanks
Nice work!!
Thanks
Thanks!
Thank you
Thank you 👋
🙏
huge plus for ollama use . thx! by the way... is it possible to tell the agents to make such a research upon my local txt files?
Yes you can integrate that via the custom tool
Thanks for sharing such a powerful tool! Been playing with it with ollama. One question though, is it just using a single model to perform multi-agent tasks? It seems to be that way when I checked my GPU resources. I’ve tried hosting multiple ollama models (emulating the GPU resources req) before. Is there a way to run multiple models as agents?
Good point, I need to explore this further about multi models multi agents . Thanks for your feedback and idea
I just saw an ollama update that demoted that.
Ollama 0.1 33: Experimental concurrency features
New concurrency features are coming soon to Ollama. They are available
• OLLAMA_NUM_PARALLEL: Handle multiple requests simultaneously for a single model
• OLLAMA_MAX_LOADED_MODELS: Load multiple models simultaneously
Really love this!!
I have a question, when i run ollama it runs the model fine on the terminal but when using AutoGen, Crewai it's very close.
It’s expected. Because local 7B parameter models are not good at tool calls and advanced tasks. It can perform only basic tasks
In the near future , you should be able to run complex tasks locally
Thanks for this video, it’s really clear to understand. But please I’m a beginner in coding and I’m finding it hard to export API in my windows terminal. What should I do? Thanks
I will create a tutorial soon to show how to do it on windows
Mervin, that’s cool! Can you imagine to include Flowise as a framework? Finally it is langchain, but with a UI.
Thanks for your suggestion, I’ll look in to this, planning for a proper UI soon
@@MervinPraison cooool!!!!
🏆
🙏
Can we use langchain or community tools in the defining tools part. How is that done?
Yes you can . In the upcoming videos will try to explain that more
@@MervinPraison Thx!
Can always use the new hermes 2 pro - llama 3 version to run this locally with ollama.
Thanks. Will test that
Where save when we type "export api" ?
To file or to memory?
In your terminal . I’ll try to create a video for beginners who are transitioning to AI , so you can learn these steps easily.
So just to be clear it doesn't work with ollama running llama 8b out of the box. 😢
Yes, not as clear as a larger model. For basic tasks it works fine
do u store the data by any means
The tool uses CrewAI and autogen , probably CrewAI will save data on vector database
So does Praison just shortcut the agent creation?
Yes correct
I really like that you got back to your tool 🤗🤩👏🙌
Thank you 🙏
I love your videos and this one in particular. I was able to install and run it all in two trials in one night. The issue was that the code expected llama3-7b while the video talked about setting OPENAI_MODEL_NAME=llama3. I only had to create one llama model with the name of llama3-7b to resolve the issue. Just to be accurate, one warning message is still persisting and I have not been able to resolve it. I hope you or one of your subscribers can propose a solution for it.
D:\AI_PraisonAI\venv\Lib\site-packages\curl_cffi\aio.py:204: UserWarning: Curlm alread closed! quitting from process_data
warnings.warn("Curlm alread closed! quitting from process_data")
Thank you so much again. This was my first fully functional agentic program. I have not had such a huge success with any other similar programs, regardless of the amount of the time I spent on them. :)
How to use RUclips for search instead of duckduckgo
pay for the youtube api search access or use a headless web crawler with selenium or puppeteer - cheerio etc