I’m excited to watch this as many times as necessary to comprehend and implement. This was a concise yet thorough walk through of a complex topic with zero fluff or dead time. Very, very well done. This video leaves me feeling empowered! I’m really looking forward to following along step by step. Thank you for this! 🙏🏼
Probably the best channel on AI LangChain actual useful applications tutorials on youtube, i always rush to see your videos, much value provided as always, thanks Jason!
Jason. DON'T STOP!!! Every single video of your is remarkable. Please continue to produce these and come up with amazing ways of using AI. Use AI to "Audit/Verify" that someone has filled up a printed form correctly (for example, with their name, address and so on) and their signature matches with another document for example an ID to verify the person completed a printed document correctly.
🎯 Key Takeaways for quick navigation: 00:00 🗣️ The video discusses how an AI research agent works, crafting AI workflows, and executing tasks. The AI agent uses large language models for its operations, with the help of tools that facilitate task completion. 00:56 🎛️ The AI agent's efficiency depends on the quality of task execution and its ability to correctly reprioritize tasks, and hence building agents for specific tasks can be more effective. 02:05 📨 An AI research agent can be used in various ways, such as triggering a workflow when someone receives an email. The agent can analyze the incoming communication, derive insights, and compile it into a specific format such as a prioritized list. 02:34 🔄 The video suggests that an AI research agent can stitch together information from different systems, providing additional decision-making support for humans. 03:18 🖥️ The host showcases how to set up a research agent, deploy it as a web service and use it in business workflows. Steps include creating tools for the agent to search the internet, scrape websites, and summarizing content. 06:07 🛠️ Details on how to create a tool for scripting website data are provided, including using functions, headers, and APIs. 07:02 📚 One of the challenges addressed in the video relates to the token limit inherent in large language models. Two strategies to handle this include the MapReduce method of summarizing content, and VectorSearch, which creates a vector embedding of a large content piece. 14:17 🌐 The host then converts the AI research agent app into a web app by using streamlit, setting page configurations, and creating text inputs. Testing is done on the research results the agent can fetch. 16:37 💻 To make the research agent more versatile, it is suggested to deploy it as a web service to be embedded in a company's business workflow. The AI research agent can be turned into an API endpoint using FastAPI. Finally, the API endpoint is tested and deployed using a cloud platform. 19:52 💻 The API endpoint developed for the research agent is live and can be called via the API. The video also proposes embedding it within the business workflow. 20:20 📩 Automating business processes using make.com is demonstrated. For example, an incoming email can trigger information extraction about sender and company, then fed into the research agent for a background check. 21:02 🎯 The video walks through the process of setting up Gmail in make.com to watch incoming emails and trigger the research agent workflow. 23:51 🧩 The integration of OpenAI is shown to extract name and company information from emails, using chat completion methods. 24:19 🌐 The extracted information is then used in the Google Cloud platform and passed on to the research agent through an API endpoint. 25:02 👨💻 The whole system is automated to run every 15 minutes, checking for emails and triggering workflows for each email. The resulting data from the AI research agent is then organized in an Airtable. 25:29 🔄 Links to other workflow templates are provided that allow the research agent to be triggered from various points like Telegram, Google Forms, ClickFunnels and even Shopify. Made with HARPA AI
Thank you very much. After going through everything, and getting this app to actually work,you've inspired me to build my first AI agent with this video, and now my mind is firing on all cylinders with the world of possibilities. Thank you for doing what you're doing.
This use case is really amazing. I am so inspired with your level of skill and knowledge in this topic. It was very informative and entertaining to watch and I can definitely see how useful this can be in day to day operation of research and business! Wow
This is amazing man! Thanks for sharing in a so detailed way 🔥🔥 especially thanks for sharing the git. And this is the best Make integration I ever seen. Used them a lot, and here’s one more use case.
You are AMAZING! Such a great video -thanks for sharing! I love how simple the the code is, and how you broke down every section and what it does. Plan on trying it with some of my own use cases soon! :)
Hey Jason, just watched your AI research agent tutorial and man, it's game-changing! Given how Botpress allows for custom chatbot training, it'd be cool to see a comparison or integration with Langchain AI agents. Keep those awesome insights coming! Cheers! 👍😃
What I think is hilarious, is that some day, AI Jason will be read and watched by an AI that is learning how to do suff. and it will say. Wow What a great site. Save that in my long term.
Amazing work Jason. Youre on fire!!! I have a few ideas/questions for your next videos. Please do a video on how to use OSS LLM such as LLama-2 to do embedding, how can we have a small scale agent, and how to do finetune/embedding for specific use case. Again, youre amazing Jason!
ohhh i like it! was planning to do my next video for embedding actually, and using LLama-2 seem make sense! Whats your use case for finetune/embedding?
@@AIJasonZ Thanks Jason, youre really on fire. Embedding is something very rare to find on youtube, and when I googled I can only find tutorials to use OpenAI embeddings. I was thinking to use it for coaching and motivator i.e. to provide emotional support when asked, so I'd like to embed philosophy and psychology such as stocism and buddha and so on.
@@AfeezAzizTV Thats a great idea! I wanted to make a life coach for people with ADHD but I never thought of adding a vector db + embeddings with works from philosophy. I think that could be a real value add. Good luck with your project!
I'm not an engineer but I know enough code to be dangerous. Having you walk us through every step is such a big help to someone like me and I'm sure many others. 非常感谢您！
@@AIJasonZ I am amazed by your tutorial's quality, value, and speed. Thank you for putting together such a thoughtful and thorough tutorial. I have several use cases that could use this methodology. My only adjustment to the process is to feed it a whitelist of sites for search as I want to limit the research to known scientific new sources but the process should hold up very well.
Building my own autonomous AI from scratch to automate red team engagement activities like social engineering and network attacks. Looking to add an autonomous research agent to build threat scenarios, so this video has been a great help!
Ooo like it notifies if a phishing attack was successful but is an inert red team attack actually? So just passively test the company's data saftey practices?
12:54 this is quite interesting. LLMs often ignores the instructions, and the way to fix it is to repeat it - or make other request. Like - one instruction per prompt, but it may be very expensive. I actually can't wait when open source models will get gpt-4 quality so I can just dedicate my own 4090 to execute multiple agents 24/7
I have been continuously learning and understanding all I can about using langchain WITHOUT OpenAI. I admit that it would be so much easier just to pay someone else for the use of their GPU's!
It's good but it doesn't actually do the search iterations. It just pulls the first results, then summarizes. (See demo on my channel). The behavior I'd like to see is manually programmed iteration: 1) Search / summarize 2) Search again to expand the ideas 3) Analyze and create a conclusion. For the 3rd step the output length should be optional. Thank you, very cool tool!
Absolutely loving this content! very informative and useful for my projects, most of the stuff you showed are stuff I have tinkered with but couldn't get working nicely, you have showed me the light!! Do you have a donation page? or a Patreon page?
Hah really appreciate your kind words, glad to hear that! 😃 I don’t have donation page or patreon, but you can subscribe to my newsletter on www.ai-jason.com; I plan to build a community soon and prob some gated content, will let you know once it is out!
This is awesome! Explained so well that as a very low coder I can run with this! I’m finding the langchain GUIs quite limited…Your videos do rock…would love to see a chainlit v stream lit v langchain chat ui video…
For example how to make them obey. They don’t always use their tools. And on a different topic, I find Zapier very confusing to use. If there a better alternative?
@@AIJasonZ getting agents to populate and work with data from Google sheets would be awesome. Basically using AI to search web and build a database. Then can read info in the database to complete other actions like emails
Really cool! Love what you've created here and it seems to work much better than the openai plugins or bing search. Would be great to see a version that minimizes external "services" like make. This seems great for personal projects, but I'm not sure how well it scales? What are your thoughts?
yea, i think there are lots of opportunities build agents that optimise specific use case! In terms of not using external services - it is definitely possible, but lots of manual work tbh, for example, connecting to google sheet api is a pain 😂 i do think it can scale, prob better than a self hosted version; but i will keep that in mind for the next video!
Hey Jason - Awesome work!! Very informaive, and clearly broken down! Q: How wold you suggest caculating the approximate total combined costs per cycle from Gmail-to-AirtableRecordEntry? Thanks again!!
@@readmarketings9061 For summation and the like, more than good enough. For code, I'd say somewhat about the same as GPT 3.5 (it may have newer information, but in my experience its less willing to give full code answers, just summarized / psuedocode). But for general writing/summation etc, its good.
I want to play around with this but am new to python and coding in general - Do you recommend that I install Python for web development on Windows, using the Windows Subsystem for Linux (WSL) or just install Python from the Microsoft store, using the basic Python3 interpreter? Sorry for the newbie question but hopefully someone can chime in so I start off on the right foot!
Excellent tutorial Jason. I have an issue with the tiktoken while its running; "PanicException: Encoder and decoder must be of equal length; maybe you had duplicate token indices in your encoder?".
So played with flowise today and it’s not giving much flexibility with serper, only parameter is URL 🤦🏻♂️Used your code for Real Estate Agent Bot 🤖 will post tomorrow
yaaaaay... for your next video can you ask it to research how to generate the maximum amount of paperclips possible please? just for a personal project I'm working on
The problem that I seen is, currently does Large Language Models (LLM) overall, have pretty short context window (Claude has about 100k context windows what is really good, but it's not nearly as smart as ChatGPT-4), and so their ability to cross examine many papers and their cross references is pretty limited, and beside that they don't seem to understand concepts, so their reasoning is quite limited too, but it's expected to improve fast, Yan LeCun and his researches are working on those problems.
@@AIJasonZ You know they're likely just going to repackage these A.I tools and call it "in-house" to fool investors like Huawei has already done, right?
I would recommend start with prompt engineer, and then start building side projects; With chatGPT as the teacher for programming you will pick up things very quickly!
Hi could you please help me with the last MAKE step you skipped that in the video and I am stuck. I have it as an HTTP post request and added the JSON but get a 422 Unprocessable Entity error in render. Would appreciate any info how you filled out the HTTP step ?
@@AIJasonZ So I put URL, Method : POST Query String: 1) Name as "query" Query String: 2) Name as "Value" Body type: Raw Content type: JSON (application/JSON) I see no options for JSON in Body type only in Content type underneath when ran I still get a 422 error. Any idea where I may be missing things I think I entered everything?
So if I just want to deploy to streamlit, I guess I wouldn’t need Step4? And can we use streamlit features like showing graph data or use Langchain to intigrate with Google sheets?
![Noob To Max With DRAGON REWORK In Blox Fruits [FULL MOVIE]](http://i.ytimg.com/vi/LnBMOinoOvA/mqdefault.jpg)
I’m excited to watch this as many times as necessary to comprehend and implement. This was a concise yet thorough walk through of a complex topic with zero fluff or dead time. Very, very well done. This video leaves me feeling empowered! I’m really looking forward to following along step by step. Thank you for this! 🙏🏼
But this guy jumps steps like if everyone knows like him.
Probably the best channel on AI LangChain actual useful applications tutorials on youtube, i always rush to see your videos, much value provided as always, thanks Jason!
Finally you released another outstanding video... I was waiting for it.
True right! I check in too much to see if he’s made a new vid😅
Jason. DON'T STOP!!! Every single video of your is remarkable. Please continue to produce these and come up with amazing ways of using AI. Use AI to "Audit/Verify" that someone has filled up a printed form correctly (for example, with their name, address and so on) and their signature matches with another document for example an ID to verify the person completed a printed document correctly.
🎯 Key Takeaways for quick navigation:
00:00 🗣️ The video discusses how an AI research agent works, crafting AI workflows, and executing tasks. The AI agent uses large language models for its operations, with the help of tools that facilitate task completion.
00:56 🎛️ The AI agent's efficiency depends on the quality of task execution and its ability to correctly reprioritize tasks, and hence building agents for specific tasks can be more effective.
02:05 📨 An AI research agent can be used in various ways, such as triggering a workflow when someone receives an email. The agent can analyze the incoming communication, derive insights, and compile it into a specific format such as a prioritized list.
02:34 🔄 The video suggests that an AI research agent can stitch together information from different systems, providing additional decision-making support for humans.
03:18 🖥️ The host showcases how to set up a research agent, deploy it as a web service and use it in business workflows. Steps include creating tools for the agent to search the internet, scrape websites, and summarizing content.
06:07 🛠️ Details on how to create a tool for scripting website data are provided, including using functions, headers, and APIs.
07:02 📚 One of the challenges addressed in the video relates to the token limit inherent in large language models. Two strategies to handle this include the MapReduce method of summarizing content, and VectorSearch, which creates a vector embedding of a large content piece.
14:17 🌐 The host then converts the AI research agent app into a web app by using streamlit, setting page configurations, and creating text inputs. Testing is done on the research results the agent can fetch.
16:37 💻 To make the research agent more versatile, it is suggested to deploy it as a web service to be embedded in a company's business workflow. The AI research agent can be turned into an API endpoint using FastAPI. Finally, the API endpoint is tested and deployed using a cloud platform.
19:52 💻 The API endpoint developed for the research agent is live and can be called via the API. The video also proposes embedding it within the business workflow.
20:20 📩 Automating business processes using make.com is demonstrated. For example, an incoming email can trigger information extraction about sender and company, then fed into the research agent for a background check.
21:02 🎯 The video walks through the process of setting up Gmail in make.com to watch incoming emails and trigger the research agent workflow.
23:51 🧩 The integration of OpenAI is shown to extract name and company information from emails, using chat completion methods.
24:19 🌐 The extracted information is then used in the Google Cloud platform and passed on to the research agent through an API endpoint.
25:02 👨💻 The whole system is automated to run every 15 minutes, checking for emails and triggering workflows for each email. The resulting data from the AI research agent is then organized in an Airtable.
25:29 🔄 Links to other workflow templates are provided that allow the research agent to be triggered from various points like Telegram, Google Forms, ClickFunnels and even Shopify.
Made with HARPA AI
Great video, Jason. I love that, as others have said, you're demonstrating how to build something actually useful.
The research agent from your link is fantastic at doing market research! The links actually work and are from reputable places. Im blown away!
Do you have to know python code before you create these apps?
just tried the AI research agent in your link, and it worked like magic! getting the research agent that dont hallucinate is so useful
Uncredibly good and easy to understand explanation. Thanks Jason!
Thank you very much. After going through everything, and getting this app to actually work,you've inspired me to build my first AI agent with this video, and now my mind is firing on all cylinders with the world of possibilities. Thank you for doing what you're doing.
Thank you for your kind words! What are the agents you try to build?
This use case is really amazing. I am so inspired with your level of skill and knowledge in this topic. It was very informative and entertaining to watch and I can definitely see how useful this can be in day to day operation of research and business! Wow
Glad it was helpful!
This is amazing man! Thanks for sharing in a so detailed way 🔥🔥 especially thanks for sharing the git. And this is the best Make integration I ever seen. Used them a lot, and here’s one more use case.
Your content is so unique and definitely one of the best channels!
You will be one of very incredible person contributing to advancement of AI. Great job. You are so authentic
Excellent content Jason, this is easily one of my favorite AI youtube channels
Bravo, very good walkthrough. I like how you pieced everything together and browserless is good for not getting blacklisted.
You are AMAZING! Such a great video -thanks for sharing! I love how simple the the code is, and how you broke down every section and what it does. Plan on trying it with some of my own use cases soon! :)
Hey Jason, just watched your AI research agent tutorial and man, it's game-changing! Given how Botpress allows for custom chatbot training, it'd be cool to see a comparison or integration with Langchain AI agents. Keep those awesome insights coming! Cheers! 👍😃
Absolutely!!!!!
What I think is hilarious, is that some day, AI Jason will be read and watched by an AI that is learning how to do suff. and it will say. Wow What a great site. Save that in my long term.
Thanks!
Dude, I focking love you! Keep up with this content! You are showing the most useful info I need for Agenst Development!
This is one of those great instructional videos to bookmark and watch multiple times. Well done!
Love your thumbnails lol, the super well explained videos are an amazing bonus :)
In the world of junk get rich quick AI content, your content is the one I never want to miss
Really amazing as usual Jason. Thank you for spending time to create the video.
Amazing work Jason. Youre on fire!!!
I have a few ideas/questions for your next videos. Please do a video on how to use OSS LLM such as LLama-2 to do embedding, how can we have a small scale agent, and how to do finetune/embedding for specific use case.
Again, youre amazing Jason!
ohhh i like it! was planning to do my next video for embedding actually, and using LLama-2 seem make sense! Whats your use case for finetune/embedding?
@@AIJasonZ Thanks Jason, youre really on fire.
Embedding is something very rare to find on youtube, and when I googled I can only find tutorials to use OpenAI embeddings.
I was thinking to use it for coaching and motivator i.e. to provide emotional support when asked, so I'd like to embed philosophy and psychology such as stocism and buddha and so on.
@@AfeezAzizTV Thats a great idea! I wanted to make a life coach for people with ADHD but I never thought of adding a vector db + embeddings with works from philosophy. I think that could be a real value add. Good luck with your project!
@@AfeezAzizTV wow it will be amazing, how can i contribute,, i want this for me.
Agreed, I've also been searching for content on open source embeddings
Loved your knowledge sharing! Really helpful! Looking forward to many more. Keep up the good work. Thanks!!!
user agent is a blessing in desguise!
very impressive. thank you for taking the time to share this awesome knowhow
I'm not an engineer but I know enough code to be dangerous. Having you walk us through every step is such a big help to someone like me and I'm sure many others. 非常感谢您！
Thanks!
Thanks mate!!
@@AIJasonZ I am amazed by your tutorial's quality, value, and speed. Thank you for putting together such a thoughtful and thorough tutorial. I have several use cases that could use this methodology. My only adjustment to the process is to feed it a whitelist of sites for search as I want to limit the research to known scientific new sources but the process should hold up very well.
My heads been spinning since watching this video. Truly interesting use case.
WE NEED MORE VIDEOS LIKE THIS MY MAN FR
First time looking at your videos Jason. Great content. Directly subscribed.
wow, great time-efficient explanation, so useful content, thank you!
Is there any chance that you will make a similar video but using some open-source model and completely free tools?
Such a brilliant video. Definitely gonna try it 🙂
Really love you videos. You are really on a different level. Great work, thanks.
Awesome bro!! Just subscribed!
Building my own autonomous AI from scratch to automate red team engagement activities like social engineering and network attacks. Looking to add an autonomous research agent to build threat scenarios, so this video has been a great help!
Ooo like it notifies if a phishing attack was successful but is an inert red team attack actually? So just passively test the company's data saftey practices?
Amazing !!!! Will be nice review an scenario with more than one AI Agent interacting.
12:54 this is quite interesting. LLMs often ignores the instructions, and the way to fix it is to repeat it - or make other request. Like - one instruction per prompt, but it may be very expensive. I actually can't wait when open source models will get gpt-4 quality so I can just dedicate my own 4090 to execute multiple agents 24/7
In a few years, GPT4 will be obsolete, like Commodore 64 obsolete 😂 you'll have something 100 times better and it'll run on a phone
I have been continuously learning and understanding all I can about using langchain WITHOUT OpenAI. I admit that it would be so much easier just to pay someone else for the use of their GPU's!
Great instructional video, thank you!
This is an amazing use case, well done.
Wow! How is it that I'm just discovering his channel now? Your videos are really amazing ❤️
Thank you!
Excellent video! 👍
that is such a cool way to use GPTs abilities! great video, and good flow
What a great video, super informative and interesting.
Really great video, thank you so much
so amazing technology😲😲.. thx to insightful video
Such an underated channel . Well done Jason
Thanks Mark, love your work!
We can probably use this for information about any stocks for the next day trading and also run some sentiments analysis on it.... This is great
Hey Jason, awesome tutorial, please do one using openAI Whisper
Great video Jason
8:00 why not train the model with the larger data? Is it more expense to train than it is to run llm chain
Excellent Conntent Jason. just well done! thank you
This was great! I’m converting this into JavaScript/ReactJS.
It's good but it doesn't actually do the search iterations. It just pulls the first results, then summarizes. (See demo on my channel). The behavior I'd like to see is manually programmed iteration: 1) Search / summarize 2) Search again to expand the ideas 3) Analyze and create a conclusion. For the 3rd step the output length should be optional. Thank you, very cool tool!
how is it working now?
Absolutely loving this content! very informative and useful for my projects, most of the stuff you showed are stuff I have tinkered with but couldn't get working nicely, you have showed me the light!! Do you have a donation page? or a Patreon page?
Hah really appreciate your kind words, glad to hear that! 😃
I don’t have donation page or patreon, but you can subscribe to my newsletter on www.ai-jason.com;
I plan to build a community soon and prob some gated content, will let you know once it is out!
This is awesome! Explained so well that as a very low coder I can run with this! I’m finding the langchain GUIs quite limited…Your videos do rock…would love to see a chainlit v stream lit v langchain chat ui video…
Amazing video, thank you so much for this!
I am very thankful to you for this!
Jason, thank you so much ! This is really amazing. I wonder if you could do a video on Agents.
Thanks Mike! Any specific topic you want me to dive deeper for agent?
For example how to make them obey. They don’t always use their tools. And on a different topic, I find Zapier very confusing to use. If there a better alternative?
@@AIJasonZ getting agents to populate and work with data from Google sheets would be awesome. Basically using AI to search web and build a database. Then can read info in the database to complete other actions like emails
Video full of 🥇 gold
Great content. Very useful stuff
Really cool! Love what you've created here and it seems to work much better than the openai plugins or bing search.
Would be great to see a version that minimizes external "services" like make. This seems great for personal projects, but I'm not sure how well it scales?
What are your thoughts?
yea, i think there are lots of opportunities build agents that optimise specific use case!
In terms of not using external services - it is definitely possible, but lots of manual work tbh, for example, connecting to google sheet api is a pain 😂 i do think it can scale, prob better than a self hosted version; but i will keep that in mind for the next video!
This is really cool, I'm working on the stuff also.
Hey Jason - Awesome work!! Very informaive, and clearly broken down!
Q: How wold you suggest caculating the approximate total combined costs per cycle from Gmail-to-AirtableRecordEntry?
Thanks again!!
Another Awesome video
Amaizing projects of AI asist. Added to my to do list
How about using Claude for summarizing? 100k tokens is pretty decent after all
No API?
@@diskomiks Ehh, yes. anthropic has api and python packages. Im using it in my app.
how about the quality when compared with gpt 3.5 turbo?
@@readmarketings9061 For summation and the like, more than good enough. For code, I'd say somewhat about the same as GPT 3.5 (it may have newer information, but in my experience its less willing to give full code answers, just summarized / psuedocode). But for general writing/summation etc, its good.
@@cholst1lately it’s been kinda regressing and allowing less uploads of pdf and csv files, as well as making up random data
I'd like to see real world testing of this and similar applications to know which issues need addressing before they are production ready.
Has anyone been using this to get stuff done? How useful and reliable is it in practice?
would request you to make more content :)
I want to play around with this but am new to python and coding in general - Do you recommend that I install Python for web development on Windows, using the Windows Subsystem for Linux (WSL) or just install Python from the Microsoft store, using the basic Python3 interpreter? Sorry for the newbie question but hopefully someone can chime in so I start off on the right foot!
Very good informatiin. Great video
Can you do a video using vector store as well.
Excellent tutorial Jason. I have an issue with the tiktoken while its running; "PanicException: Encoder and decoder must be of equal length; maybe you had duplicate token indices in your encoder?".
You are a genius :) damn... 🤯
So played with flowise today and it’s not giving much flexibility with serper, only parameter is URL 🤦🏻♂️Used your code for Real Estate Agent Bot 🤖 will post tomorrow
Nice vid. Just wondering what are the costs of running this...seems all the api's used have some form of pricing.
yaaaaay... for your next video can you ask it to research how to generate the maximum amount of paperclips possible please? just for a personal project I'm working on
Impressive work!
The problem that I seen is, currently does Large Language Models (LLM) overall, have pretty short context window (Claude has about 100k context windows what is really good, but it's not nearly as smart as ChatGPT-4), and so their ability to cross examine many papers and their cross references is pretty limited, and beside that they don't seem to understand concepts, so their reasoning is quite limited too, but it's expected to improve fast, Yan LeCun and his researches are working on those problems.
You REALLY think those are the only problems? In the end of the video this guy was suggesting we gave all our emails to open ai.
This is a masterpiece
AI jason， the number one channel to go to. I work for alibaba, and I’ve recommended this channel to jack ma already for AI literacy purpose
Say hi to jack ma for me
@@AIJasonZ You know they're likely just going to repackage these A.I tools and call it "in-house" to fool investors like Huawei has already done, right?
This is amazing. Can you make another video how this research app can combine the data from Chroma and make the research?
Hey Jason, what is your study background? What do i need to study to know all that you know?
I would recommend start with prompt engineer, and then start building side projects; With chatGPT as the teacher for programming you will pick up things very quickly!
you can also save some api calls instead of using browserless to use puppeteer, which render already has pre-installed
very useful, thanks Jason, I want more content about application of ChatGPT :)))
Have you found that 3.5 is good enough to extract accurately? I have trouble trusting it for anything beyond choosing functions
Great video!
🤯what a legend
Hi @jason ! Your vid is awesome.
1 question tho, how do you know what to fill in under GCP's Credentials: "Authorized Redirect URIs"?
Thank you. You are very good at reading and explaining code, but I would appreciate more sample runs and looking at the data
Thanks for the feedback! Is there particular area you hope I explained more?
Now it's time to automate the automation
Hi could you please help me with the last MAKE step you skipped that in the video and I am stuck. I have it as an HTTP post request and added the JSON but get a 422 Unprocessable Entity error in render. Would appreciate any info how you filled out the HTTP step ?
opps sorry just realised that this part got cut off;
So:
1. choose POST
2. Add Query string {"query":"xxxx"}
3. BodyType: JSON
@@AIJasonZ So I put
URL,
Method : POST
Query String: 1) Name as "query"
Query String: 2) Name as "Value"
Body type: Raw
Content type: JSON (application/JSON)
I see no options for JSON in Body type only in Content type underneath when ran I still get a 422 error. Any idea where I may be missing things I think I entered everything?
This is gold! ❤
How much does all this costs to set up ?
So if I just want to deploy to streamlit, I guess I wouldn’t need Step4?
And can we use streamlit features like showing graph data or use Langchain to intigrate with Google sheets?