Add AgentQL example

[jayfish0]

Sample output:

# Agent: web scraper
## Task: Scrape the main discussion topic and all the comments from this reddit post: https://www.reddit.com/r/OpenAI/comments/1h7fm9k/its_official_theres_a_200_chatgpt_pro/


# Agent: web scraper
## Thought: I will use the `query_data` tool to scrape the main discussion topic and all comments from the specified Reddit post.
## Using tool: query_data
## Tool Input: 
"{\"url\": \"https://www.reddit.com/r/OpenAI/comments/1h7fm9k/its_official_theres_a_200_chatgpt_pro/\", \"prompt\": \"Scrape the main discussion topic and all comments.\"}"
## Tool Output: 
{'discussion_topic': 'It’s official: There’s a $200 ChatGPT Pro Subscription with O1 “Pro mode”, unlimited model access, and soon-to-be-announced stuff (Sora?)', 'comments': ["As long as they're not removing the 20$ option, I'm good. It was at my upper limit as is.", 'Same. And the main improvement for o1 pro is reliability. If you can spare a few prompts, you can follow up with regular o1 telling it to test and verify, asking leading questions. This approach improves its performance a lot - just requires more human input and more prompts.', "So you're paying 180$ more for a model that can do extremely complex math problems maybe 10% better than the one you get for 20$. Not worth freaking out over tbh, this model is clearly not targeted at most people, GPT Plus is still an insanely good deal with o1.", 'Depending on company and need, the "10% better" would be worth a LOT.\n\nImagine you are a highly paid quant whose time is worth a lot and significant portion of your time is spent doing or waiting on complex math. If a tool comes by that  saves 10% of your time for 200 Dollars a month, it would be a steal if ROI on that time is worth more than 2000 Dollars a month.\n\nJust one example.\n\nFor the right kind of tools and ability to do things that are exclusive and not available to everyone, companies would easily pay many thousands a month.\n\nEdit - Not sure who downvoted this but big enterprises in finance and other such capital rich fields will not even waste their time on vendors who cannot support contracts worth 100s of thousands or millions a year. Cost and value are relative terms.', "I signed up for 200/month, I will evaluate if it's noticeably better. 200/month is nothing to an organization, a drop in the bucket...", 'Its $200 a MONTH, and geared towards scientists and researchers with huge budgets, says my ChatGPT', 'OpenAI says O1 Pro reasons for a longer duration. I reckon O1 Pro is a fine-tuned version of O1, trained with longer reasoning steps (I doubt it’s a less quantised version, as that would cost substantially more and they also explicitly say that it’s “allowed to reason for longer”).\n\nI wonder if you could get similar performance with the Plus plan’s O1 if you prompt it to “think very deeply, thoroughly, and carefully“.\n\nThe usage limit is based on the number of uses, and not tokens outputted (at least with O1  preview), so it literally won’t cost you anything more, and you could possibly get the Pro performance without the plan.', "Curious what that means for me as an ChatGPT Plus subscriber. I do understand I can't all the compute that the $200 model gets, but I would find it sad if they are also going to differentiate on features or tooling.\n\nFor example, Anthropic only offers GitHub Integration to their Claude Enterprise tier. That's just sad.", 'But 50/week limit is a bit annoying. Still, I think 200$ is insane for daily use. And from the terms of use, "Sharing your account credentials or making your account available to anyone else" is banned so you have to pay the full 200$. Maybe a $50 or even $100 plan for more o1 access limits is acceptable.']}


# Agent: web scraper
## Final Answer: 
{
  "discussion_topic": "It’s official: There’s a $200 ChatGPT Pro Subscription with O1 “Pro mode”, unlimited model access, and soon-to-be-announced stuff (Sora?)",
  "comments": [
    "As long as they're not removing the 20$ option, I'm good. It was at my upper limit as is.",
    "Same. And the main improvement for o1 pro is reliability. If you can spare a few prompts, you can follow up with regular o1 telling it to test and verify, asking leading questions. This approach improves its performance a lot - just requires more human input and more prompts.",
    "So you're paying 180$ more for a model that can do extremely complex math problems maybe 10% better than the one you get for 20$. Not worth freaking out over tbh, this model is clearly not targeted at most people, GPT Plus is still an insanely good deal with o1.",
    "Depending on company and need, the '10% better' would be worth a LOT.\n\nImagine you are a highly paid quant whose time is worth a lot and significant portion of your time is spent doing or waiting on complex math. If a tool comes by that  saves 10% of your time for 200 Dollars a month, it would be a steal if ROI on that time is worth more than 2000 Dollars a month.\n\nJust one example.\n\nFor the right kind of tools and ability to do things that are exclusive and not available to everyone, companies would easily pay many thousands a month.\n\nEdit - Not sure who downvoted this but big enterprises in finance and other such capital rich fields will not even waste their time on vendors who cannot support contracts worth 100s of thousands or millions a year. Cost and value are relative terms.",
    "I signed up for 200/month, I will evaluate if it's noticeably better. 200/month is nothing to an organization, a drop in the bucket...",
    "Its $200 a MONTH, and geared towards scientists and researchers with huge budgets, says my ChatGPT",
    "OpenAI says O1 Pro reasons for a longer duration. I reckon O1 Pro is a fine-tuned version of O1, trained with longer reasoning steps (I doubt it’s a less quantised version, as that would cost substantially more and they also explicitly say that it’s “allowed to reason for longer”).\n\nI wonder if you could get similar performance with the Plus plan’s O1 if you prompt it to “think very deeply, thoroughly, and carefully“.\n\nThe usage limit is based on the number of uses, and not tokens outputted (at least with O1  preview), so it literally won’t cost you anything more, and you could possibly get the Pro performance without the plan.",
    "Curious what that means for me as an ChatGPT Plus subscriber. I do understand I can't all the compute that the $200 model gets, but I would find it sad if they are also going to differentiate on features or tooling.\n\nFor example, Anthropic only offers GitHub Integration to their Claude Enterprise tier. That's just sad.",
    "But 50/week limit is a bit annoying. Still, I think 200$ is insane for daily use. And from the terms of use, 'Sharing your account credentials or making your account available to anyone else' is banned so you have to pay the full 200$. Maybe a $50 or even $100 plan for more o1 access limits is acceptable."
  ]
}


# Agent: sentiment analyser
## Task: Analyse the reddit comments and provide the sentiment analysis on the topic


# Agent: sentiment analyser
## Final Answer: 
1. Price: **too expensive** - The sentiment regarding price is largely negative, with many commenters expressing that $200/month is too costly, especially for personal use, whereas $20/month was considered the upper limit for some individuals. Some comments indicate that the price might be reasonable for organizations with substantial budgets, but generally, it is seen as expensive.

2. Functionality: **does not justify** - The functionality improvement from the regular to the pro version seems questionable to many commenters. Although there are mentions that the pro model offers enhanced features like greater reliability or a 10% improvement in complex math problems, the cost difference ($180 more) is not perceived as justified by the performance gains for most users.

3. Competitive sentiment: **not competitive** - The sentiment reflects that the new pricing model is not competitive with other providers or even within different tiers of OpenAI products. Some users foresee differentiation in features between current and higher-tier subscriptions, which they find disappointing. Comparisons to competitors like Anthropic suggest a concern over feature differentiation and value, diminishing the competitive appeal of the $200 model.