Grok Bot Deploys Claude Opus and Free X Data to Rival ChatGPT Agents

Peter Yang

Grok Bot eclipses rival assistants from OpenAI by pairing free X platform data with Anthropic's Claude Opus reasoning to automate multi-app business workflows.

AI-assisted summary of Grok Bot Just Got a Massive Upgrade (It’s Scary Good) — original source. Check the original for full context.

Source published: 2026-10-12

Key ideas

Grok Bot received three major platform upgrades: native and free access to X data, dedicated inbound/outbound email addresses, and Claude Opus as its new default underlying model. (0:00)
Unlike competing agent frameworks that require paid APIs or brittle browser scrapers, Grok Bot offers free access to the X firehose capped at 1,000 requests per day and 25 tweets per search. (2:49)
Grok Bot can generate proactive, scheduled email reports covering business revenue, family weekend planning, and personal health metrics without requiring manual conversational prompting. (10:06)
The platform supports multi-bot concurrency within the same thread, allowing distinct agent personas to debate operational redundancies and consolidate reporting schedules autonomously. (12:44)

Detailed notes

## Background Published by Peter Yang on YouTube on October 12, 2026, this video explores the latest upgrades to Grok Bot, which he calls his favorite personal AI agent. According to the description, the tutorial walks through ten practical use cases—including X sentiment dashboards, automated email reporting, an AI rap battle, and customer support automations—enabled by free X access, a dedicated email address, and Claude Opus 5.5 as the default foundation model.

## TL;DR Peter Yang breaks down how Grok Bot's latest upgrades—native free X access, standalone email capabilities, and Claude Opus 5.5 integration—transform it into a powerful personal operating agent. He demonstrates practical implementations across ten workflows, including competitive sentiment dashboards, content gap matrix analysis for YouTube, meeting coordination, and proactive email reports tracking business, fitness, and family life. The walkthrough also highlights advanced features like multi-bot collaboration in single threads and an autonomous community support bot, while candidly pointing out bottlenecks such as encrypted DM parsing limits and lack of full execution autonomy.

## Atmosphere & style The video feels like an engaging, over-the-shoulder workshop led by a pragmatic builder who tests AI tools against real-world operational friction rather than marketing hype. Yang's delivery is calm, clear, and methodical, seamlessly pivoting between playful creative experiments—such as an AI-generated tech-rivalry rap battle—and rigorous business automations like revenue audits and Linear ticket pipelines. Viewers experience an authentic sense of curiosity tempered by honest critique, as Yang openly flags platform guardrails and missed opportunities alongside breakthrough agent capabilities.

## Bulletin - [00:00](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=0s) Grok Bot received three major platform upgrades: native and free access to X data, dedicated inbound/outbound email addresses, and Claude Opus as its new default underlying model. - [02:49](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=169s) Unlike competing agent frameworks that require paid APIs or brittle browser scrapers, Grok Bot offers free access to the X firehose capped at 1,000 requests per day and 25 tweets per search. - [10:06](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=606s) Grok Bot can generate proactive, scheduled email reports covering business revenue, family weekend planning, and personal health metrics without requiring manual conversational prompting. - [12:44](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=764s) The platform supports multi-bot concurrency within the same thread, allowing distinct agent personas to debate operational redundancies and consolidate reporting schedules autonomously.

## Key takeaways - [00:44](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=44s) Tagging @bot on public X threads privately ingests entire conversation trees into a user's Grok Bot workspace rather than publishing replies in the public thread. - [00:44](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=44s) Personal taste profiles stored as markdown files allow Grok Bot to run recurring jobs recommending media tailored to historic preferences. - [02:49](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=169s) Free X access enables rapid creation of interactive web dashboards comparing brand sentiment, share of voice, and topic drivers against competitors. - [05:06](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=306s) Cross-platform analysis between X discussion frequency and published YouTube videos reveals content gap opportunities where discussion volume is high but video supply is low. - [07:20](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=440s) End-to-end encryption on X direct messages currently prevents Grok Bot from reading message bodies for DM triage, limiting automated filtering to sender identities and outbound links. - [08:34](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=514s) Assigning a dedicated assistant persona to Grok Bot's email address avoids the awkward social dynamic of copying an email named after the user onto scheduling threads. - [08:34](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=514s) Grok Bot currently prepares calendar responses rather than autonomously booking holds and sending confirmations, highlighting a need for greater autonomous execution. - [10:06](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=606s) Proactive email reports consolidate disparate data streams—including bank accounts, fitness apps, and funnel metrics—into daily and weekly executive summaries. - [12:44](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=764s) Users can reply directly to proactive email reports to trigger downstream tasks, such as generating structured video production tickets inside Linear. - [14:51](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=891s) Adopting Claude Opus as the default foundation model lifted Grok Bot to the top spot across independent agent intelligence benchmarks. - [14:51](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=891s) Opus enables complex creative media workflows, including pulling social context from X, generating audio via Suno, writing multi-character lyrics, and compiling motion graphic assets. - [16:57](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=1017s) Deploying specialized personas—such as support bots auditing Slack channel access and handling billing questions—is far more effective than relying on a single generalist agent. - [20:14](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=1214s) Grok Bot outperforms competitors like ChatGPT bots and Muse with Opus in terms of cleanliness, reliability, and functional versatility.

## Detailed walkthrough ### [00:00](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=0s) Grok Bot's Three Upgrades and the AI Rap Battle

Grok Bot has emerged as Peter Yang's favorite personal AI agent, receiving three major upgrades: native, free integration with X, its own dedicated email address, and Claude Opus 5.5 serving as its default model. To show off what these features can handle, Peter plans to walk through ten distinct use cases. He kicks off the video with a creative teaser: an AI-generated rap battle produced entirely by his bot, dramatizing the rivalry between Lauren from the Grok Bot team and Tibo from ChatGPT.

The battle showcases the bot's creative range with biting tech-insider banter: *"How did you know I was referring to this only? I recognize the obsession. If only you had the just put the reset in the bag, bro. What if dots weren't mid? What if goes down? 28 days."* While the production workflow behind that track is explored later in the walkthrough, Peter shifts immediately into the practical operational tools unlocked by the new release.

---

[00:44](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=44s) Tagging @bot on X for Curated Watchlists and Taste Profiles

Peter highlights the ability to summon your personal agent directly inside X threads by tagging `@bot`. To demonstrate, he pulls up a viral New York Times post listing the best television shows of the 21st century, which accumulated more than 4,700 replies. Seeking a distilled summary without having to dig through thousands of comments manually, Peter tagged both public Grok and his private `@bot` in the replies, prompting them to identify the most underrated gems and call out overrated shows.

```mermaid flowchart TD accTitle: Grok vs Grok Bot Tagging Flow accDescr: Flowchart illustrating how tagging Grok posts publicly while @bot routes privately to Grok Bot.

subgraph XPlatform["X (Twitter) Thread"] A["NYT Best TV Shows Post (4,700+ replies)"] B["User Reply: Tag @Grok and @bot"] A --> B end

subgraph PublicResponse["Public Surface"] C["Grok Public Reply<br/>Underrated: Dark, Mr. Robot<br/>Overrated: Breaking Bad"] B --> C end

subgraph PrivateAgent["Private Grok Bot Agent"] D["@bot Hook: Dispatches context to private bot"] E["Grok Bot Workspace: Detailed breakdown"] F["Taste Profile (.md): Appends Mr. Robot to watchlist"] B --> D D --> E E --> F end ```

The difference between the two tools is immediate. The public Grok assistant answers directly in the public thread, declaring that the top underrated shows are *Dark*, *Mr. Robot*, and *Westworld* (Season 1), while labeling *Breaking Bad* and *Stranger Things* as overrated. In contrast, `@bot` leaves a brief message saying, *"I'm sending this to your Grok Bot."* Over in Peter's private Grok Bot workspace, the main bot ingests the entire reply tree and generates a comprehensive, private analysis listing *The Sopranos*, *Dark*, and *Better Call Saul* as the community's favorites.

While Peter remarks that tagging `@bot` is perhaps counterproductive for public engagement on X—since other readers cannot see the actual output—it is uniquely valuable for personal productivity. For example, because he had never seen *Mr. Robot*, Peter instructed Grok Bot to add the title straight into his personal taste profile markdown file. Peter keeps an active markdown repository of all his 10-out-of-10 rated movies and series, which allows Grok Bot to run an automated weekly job recommending new films and shows tailored directly to his personal historical preferences.

---

[02:49](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=169s) Free X Access and Comparative Sentiment Dashboards

Grok Bot is currently the only personal agent on the market offering free access to the X firehose. Peter notes that despite the platform's toxicity, X remains the primary hub where artificial intelligence news breaks and professional relationships are formed. Competing agent frameworks require paid API access or brittle browser-use automations to scrape tweets. Grok Bot does operate under hard guardrails—specifically, a rate cap of 1,000 requests per day and a limit of 25 tweets per search—but Peter emphasizes that these allowances are more than generous for standard day-to-day research workflows.

Peter uses this direct access to run four distinct data use cases. The first is an automated brand sentiment analysis comparing ChatGPT Dots against Grok Bot. After initially prompting the bot to run sentiment checks on ChatGPT Dots (which generated a static graphic showing slightly negative overall user sentiment), Peter pushed the agent further, instructing it to build a fully interactive, web-based comparison dashboard.

In roughly 10 to 15 minutes, Grok Bot pulled live data from X and produced a comprehensive analytical interface: - **Timeline Filtering:** Users can filter tracking data from the exact launch date of ChatGPT Dots onward. - **Net Sentiment Tracking:** A day-by-day sentiment curve tracking positive versus negative post velocity. Overall sentiment favored Grok Bot as noticeably more positive than ChatGPT Dots. - **Share of Voice:** Despite heavy promotional pushes from OpenAI, Grok Bot maintained a commanding lead in total conversational volume on the platform. - **Platform Strengths & Weaknesses:** Comparative breakdowns detailing where each model wins and loses across key user criteria. - **Top Driver Tweets:** The dashboard surfaced top posts driving sentiment. For Grok Bot, enthusiasm was driven by its integration of Claude Opus 5.5; for ChatGPT Dots, discussions centered on always-on background agents and GPT-6 Astra. - **Underlying Post Feeds:** Users can inspect the exact tweets Grok Bot used to build its quantitative evaluations.

---

[05:06](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=306s) Tweet Performance Analytics and YouTube Content Gap Matrix

For his second use case, Peter turned Grok Bot inward to analyze the performance of his own X account between July and October. The bot produced an interactive performance audit displaying impressions, video views, and engagement metrics: - A launch post detailing Peter's "No AI Slop" evaluation scale proved to be his single highest-performing post by impressions. - The dashboard generated multidimensional breakdowns categorizing output by post format, calendar month, and conceptual theme. - A fully sortable database of historic tweets enabled Peter to identify winning content hooks and set up automated weekly advisory jobs prompting the bot: *"What post should I make more of and what post should I make less of?"*

Moving into the third use case, Peter leveraged Grok Bot to identify YouTube video topics within the AI niche by synthesizing trends from X alongside monitored YouTube competitor channels. The agent noted that content covering Anthropic's Claude ecosystem consistently overperformed, suggesting titles centered on new Claude dashboards, benchmark showdowns between Claude Haiku and Opus, and workflows embedding Claude directly inside Google Docs.

Grok Bot visualized this analysis using an interactive topic matrix comparing seven-day X post frequency against the number of published YouTube videos on that same subject. This made market gaps instantly obvious: while broad topics like Grok Bot and ChatGPT Dots were saturated with uploads, emerging topics such as "GPT-6 Intelligent UI" and "Claude Haiku" had massive discussion volume on X but very few dedicated YouTube videos. The generated dashboard included direct outbound links to outlier videos so Peter could examine the top-performing formats firsthand.

---

[07:20](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=440s) Inspecting Inbound X DMs and Current Privacy Bottlenecks

Peter's fourth test with the X integration attempted to solve the overwhelming influx of spam in his direct messages. He prompted Grok Bot to scan his past 30 days of inbound DMs and unread message requests to filter out irrelevant noise and surface high-value business opportunities or partnership inquiries.

Here, however, the bot ran into strict platform limitations: - Grok Bot reported that direct messages are end-to-end encrypted, preventing the agent from reading message bodies. - The bot could successfully read sender identities (such as inbound inquiries from users named Nate and Sebastian) and detect incoming links, but it could not parse the raw conversation text. - Peter acknowledged the necessary balance between security and functionality, but argued that when an account owner explicitly instructs their own bot to audit their inbox, the agent should have permission to process the text. He noted plans to submit direct product feedback to the development team to make private DM triage functional.

---

[08:34](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=514s) Setting Up Grok Bot's Email Address for Calendar Scheduling

The second tentpole upgrade is that Grok Bot now receives its own standalone email address. Peter offers a practical naming tip: users should avoid naming the bot's email address after themselves. Because the tool functions primarily as an executive assistant that gets CC'd onto external threads, it creates an awkward social dynamic to tell a colleague, *"Hey, let me copy in Peter to find a time for us."* Instead, Peter advises assigning the bot a dedicated virtual assistant persona—like naming it Lauren—so introductions read naturally.

Peter tested this workflow in a real scheduling scenario with an automated test contact named Zoe: 1. Zoe sent an email asking: *"Hey, Peter, are you free to catch up at 3 p.m. on Monday?"* 2. Peter looped in his bot: *"Grok, can you help us schedule 30 minutes?"* 3. Grok Bot parsed Peter's calendar, verified availability, and replied: *"You're free Monday from 3:00 to 3:30 p.m. Your only meeting this afternoon ends at 2:45. So why don't you go ahead and send this reply confirming the time with Zoe."*

While impressed that the bot cross-referenced his calendar commitments with zero friction, Peter pointed out an area needing product refinement. Rather than autonomously sending the confirmation email and booking the calendar hold, Grok Bot prepared the response and required Peter to manually confirm and hit send. In Peter's view, an ideal AI executive assistant should have the autonomy to finalize meetings directly rather than gating every action behind user approval.

[10:06](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=606s) Proactive Email Reports for Business, Family, and Health

While having Grok Bot handle calendar invites directly inside an email thread is convenient, the constant back-and-forth prompting can still feel tedious. The true game-changer of giving Grok Bot its own dedicated email address is its ability to proactively dispatch scheduled reports on a daily, weekly, and monthly basis. Instead of requiring manual check-ins, the bot continuously compiles data from integrated services and delivers digestible summaries directly to an inbox, where they are far more likely to be read than notifications buried across disparate chat windows.

```mermaid flowchart TD accTitle: Grok Bot Proactive Reporting and Task Pipeline accDescr: Pipeline showing external data sources feeding into Grok Bot, which produces email reports that trigger downstream tasks.

subgraph DataSources["Connected Data Sources"] YT["YouTube Analytics"] Bank["Mercury Bank"] Fit["Smart Scale & Fitness App"] Web["Course Funnel & Local Blogs"] end

subgraph Agent["Grok Bot Processing"] Engine["Grok Bot (Opus Engine)"] end

subgraph Outputs["Proactive Email Reports"] Biz["Business & Revenue Report"] Weekend["Friday 7 AM Weekend Planner"] Health["Health & Weight Shifts"] end

subgraph Action["Human-in-the-Loop Execution"] Reply["User Email Reply / Tag"] Linear["Linear Video Task Script"] end

YT & Bank & Web --> Engine Fit --> Engine Engine --> Biz & Weekend & Health Biz -.-> Reply Reply --> Linear ```

These automated briefs cover three major pillars of daily life and work: - **Business Operations:** Connects directly to YouTube metrics, Mercury bank accounts, website analytics, and course platform funnels. Following the recent launch of a personal AI course, the bot automatically aggregates conversion rates, top-of-funnel reach, total revenue, and running expenses to reveal which business growth levers need attention. - **Weekend Family Planning:** Runs automatically every Friday at 7:00 AM. It analyzes local community blogs and event listings around specified neighboring towns to suggest family-friendly outings, ensuring quality time with the kids. It also references an established personal taste profile to recommend weekend movies and television series. - **Personal Health Tracking:** Ingests biometric data from a smart scale alongside activity logs from a custom "vibe-coded" fitness tracking app to monitor weight shifts, strength training consistency, and workout volume.

---

[12:44](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=764s) Turning Email Reports into Linear Tasks and Multi-Bot Collaboration

Beyond static reading, these proactive emails function as an interactive launchpad for complex workflows. In a newly configured market research email that combines live X sentiment and YouTube trends, the workflow can be triggered simply by hitting reply. By emailing instructions back to Grok Bot, it was assigned to generate a complete video production ticket inside Linear, the project management tool used to manage all video production pipelines.

Within moments, Grok Bot created a fully structured ticket for a proposed video on the GPT-6 intelligent UI, complete with an audience hook, production outline, and supporting research notes. While any creator still needs to perform a manual editorial pass to confirm the concept's viability before filming, delegating the initial drafting and workspace organization eliminates massive amounts of administrative friction.

A standout capability that separates Grok Bot from competing platforms like Chatbox or Muse is support for multi-bot collaboration within the exact same conversation thread. When a YouTube Producer bot and an Advisor bot began generating overlapping market research briefs, both bots were brought together in a single thread to resolve the redundancy. The bots actively discussed options—debating whether to maintain two cleanly split reports or consolidate everything into a single, compact Friday morning dispatch—before settling on a unified market research schedule.

---

[14:51](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=891s) Upgrading to Claude Opus: Intelligence Benchmarks and Rap Battles

The most consequential platform upgrade is that Grok Bot now defaults to Anthropic's Claude Opus as its core foundation model rather than Grock 4.7. While earlier proprietary Grok models held their own, independent intelligence benchmarking conducted by Matt Van Horn demonstrated that Grok Bot vaulted to the number one spot across intelligence rankings immediately after adopting Opus.

Opus unlocks creative and technical workflows that earlier models could not manage, including motion graphics generation, multimedia direction, and nuanced pop-culture satire. To demonstrate this capability, Grok Bot was tasked with transforming a real-world tech feud on X between user Lauren (@potato) and Thibault from ChatGPT into an aggressive 30-to-60-second animated rap battle.

The end-to-end production pipeline highlights the creative agility of the Opus integration: - Opus pulled context, insult exchanges, and memorable arguments directly from their public tweet threads on X. - It connected to Suno (suno.com) to generate free, custom audio stems and rap vocals. - It wrote dedicated battle lyrics, iterating through drafts to ensure Lauren was assigned a distinct female vocal track while Thibault was matched with an opposing male voice. - It designed visual fight cards, generated an image of Grok Bot clashing with competing AI avatars, and structured motion graphic sequences before compiling the final video.

---

[16:57](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=1017s) Autonomous Support Bots and the Power of Specialized Personas

Putting agents into production requires moving beyond open-ended conversational novelties and assigning models explicit business responsibilities. Within the paid Slack community run through Behind the Craft, an autonomous agent dubbed the "Behind the Growth" bot handles day-to-day operations and customer support to reclaim hours of manual administrative labor each week.

The bot conducts automated daily sweeps across the community Slack to process operational requests without human intervention: - Auditing active member rosters to verify that only paying subscribers have access to private channels. - Triaging customer billing questions, resolving account email mismatches, and processing refund requests or payment disputes. - Drafting policy-compliant responses for unique inquiries. When a prospect emailed asking if a lifetime membership plan was available instead of standard annual billing, the bot drafted an polite response clarifying that only annual memberships exist while taking note of the feature request, pausing only to ask for final approval before hitting send.

This persona-driven structure fundamentally reframes how AI fits into daily operations. Instead of dealing with disjointed, generalist chat windows, the user interacts with dedicated digital employees: a customer support agent for community management, a YouTube Producer for content prep, a Loving Husband persona for coordinating family purchases and bookings, a Weekend Planner for itineraries, and a Health Coach for personal wellness.

---

[18:48](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=1128s) Full Capabilities Recap

The latest wave of platform enhancements turns Grok Bot into an end-to-end autonomous assistant ecosystem, anchored by several core operational features:

- **Universal Tagging on X:** Calling `@bot` anywhere on X instantly triggers targeted research, thread extraction, or data analysis inside the Grok Bot dashboard. - **Unrestricted X Data Access:** Native integration allows real-time brand sentiment tracking, retrospective viral post breakdown, cross-platform audience overlap analysis between YouTube and X, and future capabilities around automated DM triage. - **Dedicated Inbound/Outbound Email:** Custom email routing enables automated meeting scheduling, cross-platform research generation, and scheduled proactive reporting across business, fitness, and family life. - **Multi-Bot Concurrency:** Multiple distinct agent personas can inhabit the same workspace thread to debate strategies, eliminate duplicate workflows, and consolidate team outputs. - **Default Opus Architecture:** Incorporating Claude Opus as the base engine provides best-in-class reasoning for complex coding, media production, and autonomous multi-step support workflows.

[20:14](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=1214s) Final Verdict and Comparison with Other Agents

Wrapping up his walkthrough of the customer support bot built for his paid community, Peter emphasizes that setting up a similar workflow is straightforward for anyone looking to automate customer inquiries. Delivering his honest assessment, Peter stresses that he always strives to provide "the real deal": right now, Grok Bot has legitimately become his favorite personal agent.

Comparing it to competing platforms across the AI landscape, Peter notes that Grok Bot feels cleaner and significantly more reliable than ChatGPT bots, while simultaneously proving more capable and powerful than tools like Muse with Opus. He strongly encourages viewers to test out Grok Bot's upgraded feature set for themselves, signing off with a call to like and subscribe for more practical, no-nonsense AI tutorials in the future.

- **Community Support Implementation:** Viewers can replicate the exact setup demonstrated to automate customer service inquiries across paid communities and memberships. - **Agent Comparison:** Grok Bot is rated as cleaner and more reliable than ChatGPT's bots, as well as more powerful than Muse with Opus.

## Notable quotes > [00:00](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=0s) How did you know I was referring to this only? I recognize the obsession. If only you had the just put the reset in the bag, bro. What if dots weren't mid? What if goes down? 28 days.

> [00:44](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=44s) I'm sending this to your Grok Bot. — @bot

> [05:06](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=306s) What post should I make more of and what post should I make less of? — Peter Yang

> [08:34](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=514s) Hey, let me copy in Peter to find a time for us. — Peter Yang

> [08:34](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=514s) Hey, Peter, are you free to catch up at 3 p.m. on Monday? — Zoe

> [08:34](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=514s) Grok, can you help us schedule 30 minutes? — Peter Yang

> [08:34](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=514s) You're free Monday from 3:00 to 3:30 p.m. Your only meeting this afternoon ends at 2:45. So why don't you go ahead and send this reply confirming the time with Zoe. — Grok Bot

> [20:14](https://www.youtube.com/watch?v=8hhFOmxEbT0&t=1214s) the real deal — Peter Yang

## Mentioned Peter Yang, Grok Bot, Grok, X, Claude Opus 5.5, Claude Haiku, Anthropic, ChatGPT, ChatGPT Dots, GPT-6 Astra, GPT-6 Intelligent UI, Grock 4.7, Lauren (@potato), Thibault, Matt Van Horn, The New York Times, Dark, Mr. Robot, Westworld, Breaking Bad, Stranger Things, The Sopranos, Better Call Saul, YouTube, Linear, Mercury, Suno, Behind the Craft, Slack, Chatbox, Muse

## Video description Copy my best AI workflows to save time and automate busywork: https://www.behindthecraft.com 🤝 Subscribe to my practical AI newsletter read by 150K+ professionals: https://creatoreconomy.so

Grok Bot is my favorite personal agent right now, and it just got 3 big upgrades: free access to X, its own email address, and Claude Opus 5.5 as its default model. In this tutorial, I walk through 10 real use cases, from X sentiment dashboards and YouTube research to proactive email reports, a rap battle made with Opus, and a bot that runs customer support for my paid community, plus where Grok Bot still falls short.

What I cover: (00:00) Grok Bot's 3 big upgrades (and a rap battle my bot made) (00:44) Tag @bot on X to put your Grok Bot to work (02:49) Free X access: sentiment analysis, tweet analytics, and YouTube ideas (08:34) Grok Bot gets its own email address (10:21) Proactive email reports for my business, family, and health (14:51) Opus 5.5 is now the default model, and how I made the rap battle (16:57) A support bot that runs my paid community (18:48) Recap and my honest take

Where to find me: X: https://x.com/petergyang LinkedIn: https://www.linkedin.com/in/petergyang/

Subscribe to this channel - more tutorials coming soon!

[Source](https://www.youtube.com/watch?v=8hhFOmxEbT0)

中文简报

Grok Bot整合Claude Opus与免费X数据，对标ChatGPT智能体 Grok Bot将免费的X平台数据与Anthropic的Claude Opus推理能力相结合，用于自动化多应用业务工作流，从而超越OpenAI的竞品助手。 Peter Yang详细分析了Grok Bot的最新升级——原生免费访问X、独立电子邮件功能以及集成Claude Opus 5.5——如何将其转变为强大的个人操作智能体。他演示了十个工作流中的实际应用，包括竞品情绪仪表盘、针对YouTube的内容差距矩阵分析、会议协调，以及追踪业务、健身和家庭生活的主动电子邮件报告。该解析还重点介绍了高级功能，例如单线程中的多机器人协作以及自主社区支持机器人，同时坦率地指出了加密私信解析限制和缺乏完全执行自主权等瓶颈。 Grok Bot迎来了三项重大平台升级：原生且免费访问X数据、专用的收发电子邮件地址，以及采用Claude Opus作为其新的默认底层模型。 (0:00) 与需要付费API或不稳定的浏览器爬虫的竞争对手智能体框架不同，Grok Bot提供对X数据流的免费访问，上限为每天1,000次请求，每次搜索25条推文。 (2:49) Grok Bot无需人工对话提示，即可主动生成涵盖业务收入、家庭周末计划以及个人健康指标的定期电子邮件报告。 (10:06) 该平台支持在同一线程中进行多机器人并发协作，允许不同的智能体角色自主探讨运营冗余并整合报告日程安排。 (12:44)

Reading summary

Peter Yang reports that Grok Bot has become his preferred personal AI agent following major upgrades that introduce native X integration, a dedicated email address, and Claude Opus as its default model. Real-Time X Scraping and Market Analytics (2:49) Grok Bot offers direct access to X's data feed without requiring external scrapers or paid APIs, subject to rate caps of 1,000 requests per day and 25 tweets per search. Users can summon the agent within public X threads by tagging @bot, which silently routes the thread into a private workspace rather than replying publicly. Yang used this mechanism to summarize lengthy discussions and log recommendations into a personal markdown taste profile, which powers automated weekly media watchlists. Yang also deployed the bot to conduct market research, building a functional comparative sentiment dashboard in 10 to 15 minutes that tracked conversation volume and net sentiment between Grok Bot and ChatGPT Dots. In addition, the integration enabled an internal performance audit of Yang's own account metrics, as well as a cross-platform topic matrix that identified content gaps where subjects had high discussion volume on X but low video saturation on YouTube. Email Integration and Proactive Asynchronous Reporting (10:06) Assigning Grok Bot a dedicated email address allows it to deliver scheduled reports directly to an inbox instead of requiring manual check-ins. Yang set up proactive dispatches across three areas: business metrics (aggregating Mercury bank data, YouTube analytics, and course funnel conversions), family logistics (a Friday morning scan of local community blogs for weekend events), and personal health (monitoring biometric data from a smart scale and a fitness app). These automated emails also serve as operational triggers. By replying directly to a market research summary, Yang directed Grok Bot to create a structured video production ticket inside Linear, complete with an outline and audience hook. Grok Bot also supports multi-bot concurrency, allowing separate specialized agents—such as a YouTube Producer and an Advisor bot—to communicate within a single thread to eliminate redundant tasks and consolidate reporting schedules. Claude Opus Integration and Specialized Personas (14:51) Grok Bot replaced Grock 4.7 with Anthropic's Claude Opus as its default foundation model, a change Yang notes elevated the bot to the top spot in independent intelligence benchmarks conducted by Matt Van Horn. This reasoning capability supports complex multimodal pipelines; to demonstrate, Yang had the bot research an online dispute between tech workers on X, write battle lyrics, generate audio tracks through Suno, and assemble an animated rap battle video. In daily operations, Yang structures the agent into dedicated roles rather than using a single generalist assistant. In his paid Behind the Craft community, a dedicated Slack bot audits member rosters against active subscription lists, triages billing and refund requests, and drafts customer support emails for human approval before sending. Security Boundaries and Current Automation Bottlenecks (7:20) Despite the bot's expanded functionality, Yang points out several operational frictions and platform guardrails. When attempting to triage inbox spam and surface partnership inquiries on X, Grok Bot was blocked by end-to-end encryption; the agent could identify sender handles and outbound links, but could not read message bodies to evaluate content. Administrative limitations also persist in calendar management. While Grok Bot successfully cross-referenced Yang's calendar availability to find an open meeting slot, it only drafted the proposed response and required Yang to confirm and send the email manually. Yang argues that an optimal executive assistant should possess the autonomy to finalize calendar holds and confirmations directly. Yang evaluates Grok Bot as cleaner and more reliable than ChatGPT bots and more capable than tools like Muse with Opus. However, realizing full executive delegation remains dependent on resolving platform hurdles, particularly enabling permissioned access to encrypted direct messages and allowing the agent to finalize calendar invites autonomously.

Source quotes

[0:00] How did you know I was referring to this only? I recognize the obsession. If only you had the just put the reset in the bag, bro. What if dots weren't mid? What if goes down? 28 days. [0:44] @bot: I'm sending this to your Grok Bot. [5:06] Peter Yang: What post should I make more of and what post should I make less of? [8:34] Peter Yang: Hey, let me copy in Peter to find a time for us. [8:34] Zoe: Hey, Peter, are you free to catch up at 3 p.m. on Monday? [8:34] Peter Yang: Grok, can you help us schedule 30 minutes? [8:34] Grok Bot: You're free Monday from 3:00 to 3:30 p.m. Your only meeting this afternoon ends at 2:45. So why don't you go ahead and send this reply confirming the time with Zoe. [20:14] Peter Yang: the real deal

Original transcript

[0:00] Grogbot is my favorite personal agent [0:02] right now, and it just got better with [0:04] 3X integration, its own email address, [0:08] and Opus as its default model. In this [0:11] video, I'm going to show you 10 amazing [0:13] use cases with these new features, such [0:15] as this epic rat battle that my bot [0:18] made, capturing the feud between Lauren [0:20] from the Grobbot team and Tibo from Chad [0:23] GPT. [0:24] >> How did you know I was referring to [0:26] >> this only? I recognize the obsession. If [0:29] only you had the just put the reset in [0:32] the bag, bro. What if dots weren't mid? [0:34] >> What if goes down? [0:37] >> 28 days. [0:40] >> All right, I'll show you how I made that [0:42] rap battle later in this video. But [0:44] first, let's cover some super practical [0:46] use cases, starting with being able to [0:49] tag bot directly on X to get your bot [0:52] working. Here's a tweet from the New [0:54] York Times listing the best TV shows of [0:57] the century. If you scroll down here, [0:59] there are over 4,700 replies. So, what I [1:02] did was I tagged both Grock and Bot in [1:05] the replies asking to recommend the top [1:08] underrated shows and also talk about [1:11] which shows are overrated. Let's take a [1:13] look at Grock first. Grock replied [1:15] publicly here and it said that the top [1:18] underrated shows are Dark, Mr. Robot, [1:20] Westworld, season 1, and so on. And the [1:23] most overrated shows are Breaking Bad, [1:25] Stranger Things, and more. [1:28] So while Grock replies publicly, if we [1:31] go back here and we take a look at bot, [1:34] bot says, I'm sending this to your [1:35] Grockbot. So we can pull out what the [1:38] responses are asking for. And now let's [1:40] go over to my Grockbot. Okay, so you can [1:43] see here that my main bot was triggered [1:45] by the app bot mentioned on X and it's [1:47] following up on my request, right? And [1:50] here it's replying with a more detailed [1:52] reply saying these are the most [1:53] recommended shows, The Sopranos, Dark, [1:55] Better Call Saul, and these are the most [1:58] overrated shows, Stranger Things, [1:59] Breaking Bad, and so on. Now, I don't [2:02] think being able to tag appbot on X is [2:04] actually good for X itself because no [2:08] one else can see the replies here. But I [2:10] think it's extremely useful for [2:12] follow-up requests like this. Let's say [2:15] I didn't watch Mr. Robot before and I [2:17] want to add Mr. robot to my watch list [2:19] and taste profile. And Grockbot is able [2:21] to go ahead and do that, adding to my [2:24] markdown file here. By the way, if you [2:26] love movies and TV shows, I highly [2:29] recommend setting up a taste profile [2:30] markdown file where you keep track of [2:32] all your rated movies. So, here's an [2:35] example. These are all the 10 out of 10 [2:37] rated TV and movies that I added to my [2:40] taste profile. This way, I can ask [2:42] Grockbot to set up, for example, a [2:43] weekly job to recommend me new movies [2:46] and TV shows to watch based on my [2:48] existing tastes. It's just extremely [2:50] helpful. Okay, next, Grockbot is now the [2:53] only personal agent that has free access [2:56] to X. And this is obviously a game [2:59] changer because X is where all the AI [3:01] news breaks, where I made a lot of new [3:03] friends and connections. And as toxic as [3:06] this platform is, it's just very useful. [3:09] Now, there are limits. Grockbot can only [3:11] pull from X a thousand times a day, and [3:14] each time it can only search up to 25 [3:17] tweets, but I think for the regular [3:18] user, these limits are more than enough. [3:21] Okay, now let's walk through four real [3:24] use cases that are extremely useful with [3:26] your X data. The first one is you can [3:28] ask your bot to do a sentiment analysis [3:31] on your brand and even compare it with [3:33] another brand. Here I asked my Xbox to [3:37] do a sentiment analysis on chat GPT dots [3:41] and here it came up with this image. [3:43] Zooming here and you can see that the [3:46] sentiment overall is slightly negative. [3:49] But I didn't stop there. I then asked [3:51] Grothbot to make me an interactive [3:54] website and compare chatgbts sentiment [3:57] against Grothbot. And let's take a look [4:00] at what I made. Awesome. So here is the [4:02] sentiment comparison between Chad Gibb [4:05] dots and Grock. You can see here that [4:07] Grockbot is more positive overall in [4:10] sentiment than CHB dots. But I was blown [4:13] away by how robust this dashboard is. So [4:15] I can filter down to since dots [4:17] launched. I can see the net sentiment [4:19] over time by day on X. I can also see uh [4:23] share of voice. It seems like despite a [4:25] lot of promotion from OpenAI, Grockbot [4:28] is still by far the highest share of [4:30] voice. And down here I can see where [4:32] each platform wins and loses on [4:35] different criteria. And even down here I [4:37] can see for the X tweets that mention [4:40] each product what are the top tweets. So [4:42] for Grockbot it's the fact that it's now [4:44] using Opus 5.5. And for chat GPD dots [4:47] it's always on agents and GPD6 Astra. [4:50] This is like a wealth of information [4:52] that Grockbot was able to pull from X [4:55] data in around 10 to 15 minutes. And [4:59] down here I can also see the actual post [5:01] that is pulling data from. So this is [5:04] pretty incredible. But let's not stop [5:06] there. The next use case is asking [5:08] Grockbot to pull data on the performance [5:11] of my own tweets. And you can do this [5:13] for your own account too. So here is a [5:16] video tweet performance of my account [5:18] from July to October. And I can see [5:20] impressions and everything else. And [5:22] here's a handy chart that shows which [5:24] tweets are getting the most impressions. [5:26] So, for example, I did a launch post [5:28] about my no AI slop scale that seems to [5:30] have received the most impressions. And [5:32] down here is a breakdown by format. I by [5:35] month and by theme. And again, here is [5:38] the list of posts that I made. And I can [5:40] sort it by how many impressions I got [5:42] and how many video views I got to [5:44] understand what is doing well and what's [5:46] not doing well. And you can easily [5:48] imagine that I can set this up to [5:49] refresh, let's say, every week, and [5:52] follow up with questions about, okay, so [5:54] what post should I make more of and what [5:55] post should I make less of to help me [5:57] optimize my own content. So, I just find [6:00] this stuff super useful. All right, use [6:03] case three is getting inspiration for [6:06] YouTube videos to make in my niche. So [6:08] here it's pulling from both X data and [6:11] also a bunch of channels that I'm [6:13] monitoring on YouTube and it's giving me [6:14] specific recommendations for videos to [6:16] make. So right away it's recommending I [6:19] make videos about the new cloud [6:21] dashboards testing cloud haiku versus [6:23] opus cloud now works inside Google docs [6:26] and so on. Basically videos that mention [6:27] cloud tend to do well. And down here I [6:30] was really kind of blown away by this [6:31] topic matrix. So what it does is it [6:33] compares how many channels have made [6:36] YouTube videos about a topic versus how [6:38] many exposts about a topic there has [6:41] been over the last seven days. I can see [6:43] here that videos about Grockbot and Chib [6:46] dots and stuff a lot of channels are [6:47] making this stuff but it's finding [6:49] opportunities in GPT6 intelligent UI [6:52] cloud haiku like this kind of stuff [6:54] where there isn't a lot of videos yet. [6:56] So, it's really useful to help me [6:59] identify gaps for videos to make. And [7:02] down here is a list of actual outlier [7:04] videos. And if I can click any of the [7:06] links, I can actually go watch the video [7:08] myself. So again, you can set up this [7:10] kind of research report for any kind of [7:12] trend or market analysis that you want [7:14] to do. And here is combining both X data [7:17] and also YouTube data. Awesome. The last [7:20] use case that I want to highlight with X [7:22] is looking at your DMs. You know, if [7:26] you're like me, you just get a ton of [7:27] spam DMs, a lot of random people DMing [7:30] you, and it's hard to separate the [7:32] signal from the noise. So, I asked [7:35] Rockbot, can you look at my past 30 days [7:37] of inbound DMs, and is there any [7:39] interesting things like business [7:40] opportunities and so on so forth? And in [7:43] this case, it didn't quite do its job. [7:45] So, it said that your DMs are encrypted. [7:47] So, I can tell who is DMing you, like [7:50] Nate and Sebastian, but I can't actually [7:52] tell you what the content of the DM is. [7:55] And down here, I asked it to look at my [7:58] message requests because I also get a [8:00] lot of DMs from people who I'm not [8:01] connected with. And it found a bunch of [8:04] links here, but again, it wasn't able to [8:06] pull the contents. So, I understand like [8:08] the whole privacy and security thing, [8:09] but I think if the user is asking you to [8:10] look at your DMs, like Grockbot should [8:13] be able to do this analysis. So, I think [8:15] I'm going to give the team some feedback [8:16] about this. Awesome. So, that was a [8:18] quick demo of Grockbot using X data. And [8:21] again, Grockbot is the only personal [8:24] agent that has free access to X data. [8:27] Everybody else has to use X API or maybe [8:29] use browser use to try to figure this [8:30] out. So, this is a big advantage that [8:32] Grabbot has. But now, let's move on to [8:35] the next feature, which is Grathbot now [8:37] has its own email. And my pro tip here [8:41] is that it doesn't actually make sense [8:43] to call your Grockbot email after [8:45] yourself because the most common use [8:48] case is you got to think about Grockbot [8:49] as kind of an executive assistant and [8:51] you want to copy them in, for example, [8:52] to help you schedule a meeting with a [8:54] co-orker or a friend. And it really [8:56] doesn't make sense to say, "Hey, let me [8:58] copy in Peter to find a time for us." So [9:00] my advice is naming Grabbot something [9:02] else like a virtual assistant name or [9:04] something so that you can be like, "Hey, [9:06] I'm copying Lauren or someone else in." [9:09] So, speaking of which, I tested this use [9:11] case. By the way, Zoe is like another [9:12] bottom of mine, but basically I I tested [9:14] Zoe emailing me saying, "Hey, Peter, are [9:16] you free to catch up at 3 p.m. on [9:18] Monday?" And I added my Grockbot email [9:21] into this chain. I said, "Grock, can you [9:23] help us schedule 30 minutes?" And what [9:26] happened next was that this triggered my [9:29] email bot to say, "Hey, you're free [9:31] Monday from 3:00 to 3:30 p.m. Your only [9:34] meeting this afternoon ends at 2:45. So [9:37] why don't you go ahead and send this [9:38] reply confirming the time with Zoe. Now [9:42] my hope is actually that Grockbot [9:44] automatically schedules the meeting [9:46] because it recognized that I'm free [9:47] during the time, but instead it's asking [9:48] me to confirm and send the email myself. [9:51] So I think there's some room for [9:52] improvement here. I think a true virtual [9:54] assistant doesn't ask the user for [9:56] approval each time and just does things. [9:59] But I think this is still pretty useful. [10:01] So, if I hit send here, then what's [10:04] going to happen is Grabbot is going to [10:06] reply in that email thread. And again, [10:09] if I hit yes, add the invite, it is [10:11] going to add the invite for me. So, [10:13] overall, I think this is great. I just [10:15] wish there was a little bit less [10:16] prompting from Grockbot to the user and [10:18] Grobot just handled this stuff [10:19] automatically. But honestly, I think the [10:22] real unlock with Grockbot having it own [10:24] email is that it can now proactively [10:27] send me the schedule reports on a daily, [10:30] weekly, and monthly basis. So, let me [10:33] show you some examples of reports that I [10:35] have my Grockbot send me. On the left is [10:37] a report with a overview about my [10:39] business and how it's doing. You know, [10:41] recently I launched a new personal AI [10:43] course. Seems to be doing well. Here's [10:45] how my funnel is doing. Here's how my [10:47] YouTube is doing. And finally, some [10:50] other information about my revenue and [10:52] expenses. And it's able to send this [10:54] report on the left because I connected [10:56] Grockbot to my YouTube data, to Mercury, [10:59] my bank, and also to my website and [11:03] course metrics. So, this is just like [11:05] super useful to keep me on track of how [11:08] my business is doing and what other [11:09] levers I should pull to grow my [11:11] business. In the middle here is a report [11:14] that's a weekend activities report. What [11:16] it does is it analyzes local blogs and [11:19] event websites and it recommends a list [11:21] of activities for me and my family to do [11:24] on the upcoming weekend. I find it super [11:26] useful just to make sure that I'm [11:28] spending quality time with my kids. And [11:30] on the right is a health report that I [11:33] have my rockbot send me based on data [11:36] from my smart scale and also from a [11:38] fitness app that I have vibe coded. So [11:40] it keeps track of my workouts and how my [11:43] weight is shifting. These are some of [11:45] the most insightful emails that I get [11:46] about my business and life and I highly [11:49] recommend that you do the same. [11:51] Actually, let me just quickly show you [11:52] how this stuff works. Okay, so I have a [11:55] weekend planner bot that sends me the [11:58] weekend fund report. You can see that [12:00] there is a report that runs every Friday [12:02] at 7 a.m. And it also sends me the stuff [12:04] in chat here, too. And the reason I like [12:06] email is because I tend to read my [12:08] emails more than all these different [12:09] chat threads. But let's scroll down here [12:11] and let's see how this report works. [12:14] Okay, so here is the actual prompt that [12:17] drives the report. You can see here that [12:19] it has specifications about what the [12:20] title is, how to find activities nearby. [12:23] Here's the nearby towns that I visit. [12:26] And also in the same report, it sends me [12:27] a list of movies and TV shows to watch [12:29] based on the taste profile that I [12:31] mentioned earlier. But of course, I [12:34] didn't actually write this prompt from [12:36] scratch. I worked with my weekend [12:38] planner bot to come up with this and to [12:40] iterate on it until it got to something [12:42] that I really liked. So I recently just [12:45] added a new report to my crockbot called [12:47] market research. And here is looking at [12:50] both X data and YouTube data, the same [12:52] dashboard that you saw earlier, but [12:54] instead it's sending to me as an email. [12:57] And the brilliant thing about all this [12:58] is that I can actually just reply to [13:01] this email and tag Rockbot to Cal to [13:04] start doing work. Okay, so what I did [13:06] was I added my Grockbot email and I [13:09] asked it to draft a linear ticket with [13:11] the hook outline for the chatbt [13:14] intelligence video idea and drafted in [13:17] linear which I use to manage all my [13:20] video scripts. And lo and behold, when I [13:23] go to linear, you can see that my [13:24] Grockbot has drafted the GPD6 [13:27] intelligent UI video. There's a hook, [13:29] there's an outline, and everything else [13:31] based on the research that I did. Now, [13:33] of course, I'm not just going to make [13:34] this video. I have to do a manual pass [13:36] and make sure this is a good idea. But [13:38] you can see how this saves me just a ton [13:39] of time. So, another thing that I really [13:42] love about Grogbot is that you can [13:44] actually add multiple bots to the same [13:46] chat thread, which I don't think chat [13:49] dots or muse or any of the other [13:50] alternatives can do. So, for example, [13:53] here I added my YouTube producer and my [13:55] advisor bot to the same chat thread and [13:58] both were sending me emails that were a [14:00] little bit duplicative that had YouTube [14:02] market research information. So, I asked [14:04] them like, hey, can we consolidate stuff [14:06] into one report and they went back and [14:09] forth on it? and YouTube producer [14:11] eventually suggested, hey, you can [14:13] either split the reports cleaning and [14:14] get two reports or you can consolidate, [14:17] get a short market read on Friday and [14:20] you can keep everything in one Friday [14:22] email. And then we went back and forth [14:26] before the two boss decided to send me a [14:28] separate market research email that I [14:30] shared with you up here. So, quick plug. [14:34] If you want the prompts to set up all [14:37] these proactive workflows for yourself, [14:39] either in Grockbot or another AI tool, [14:41] check out behindthecraft.com [14:43] where I have a full course that walks [14:45] through how to set up each workflow on [14:48] the left here. That's [14:49] behindthecraft.com. [14:51] Okay. Okay, the next feature I want to [14:52] walk through is probably the most [14:53] important one, which is that Grockbot [14:56] now defaults to the best model in the [14:59] market and right now that is Opus from [15:02] Anthropic. [15:04] So, previously Grockbot was using Grock [15:06] 4.7 as a default and the Grock models [15:10] aren't necessarily bad, but they're [15:11] nowhere near as smart as Opus. My friend [15:14] Matt Van Horn actually did an [15:15] intelligence test. And previously [15:17] Grockbot was actually not super high in [15:20] intelligence, but the upgrade to Opus [15:23] made Grockbot number one in intelligence [15:26] as you can see from Matt's latest test. [15:29] All right, so let me return to the rap [15:31] battle video and how I made it in [15:33] Grockbot. If you know Opus, you know [15:35] that it's incredible at making motion [15:38] graphics and editing videos. So, [15:40] basically, I asked Mike Grockbot, I want [15:43] you to look at the feud between Lauren, [15:45] whose handle is Potato, and Tibo from [15:47] Chad GBT, and make an incredible, [15:49] hard-hitting 30 to 60 second video of [15:53] their feud. And here's some links to [15:55] their best tweets back and forth, and [15:58] also add some music, too. And then, [16:01] Grockbot with Opus got to work. I asked [16:04] it to use Sunno. You can check it out at [16:06] sunno.com. It's another awesome AI tool [16:09] for making music for free to make the [16:11] actual rap lyrics at the battle. And you [16:14] can see down here that Grockbot got to [16:16] work. It even made an image of Grockbot [16:19] and Dots fighting each other. And it [16:21] also wrote the lyrics here for me to [16:24] review. I then suggested that it show [16:26] profiles of Lauren and Tibo themselves. [16:30] And there was a bunch of iteration on [16:32] actually the music, making sure that [16:34] Lauren was a female voice and the table [16:36] was the male voice rapping against each [16:37] other. But basically, thanks to Opus, it [16:40] didn't take too long before Grockbot [16:43] produced this pretty awesome rap battle [16:45] here. Unfortunately, Potato and Tibo [16:48] have since made up. They're no longer [16:49] fighting with each other. But thanks to [16:51] videos like this and more, their feud [16:54] will always be remembered on X. Okay, I [16:57] want to close by sharing a real bot that [17:00] I'm using to perform a job that saves me [17:02] at least a couple hours every week. So, [17:05] I have a Slack community with other AI [17:07] builders and we trade notes and [17:10] everything and you can get into the [17:11] community again at behindthecraft.com. [17:13] But basically, I have a behind the [17:16] growth bot that does customer support [17:19] for the community. So, the job of this [17:21] bot is to support my behind the craft [17:24] community. It handles refunds, [17:25] cancellations, billing questions, and [17:28] also making sure that only paid users [17:30] are in my Slack. And every day, I have [17:34] it run a daily support sweep of my Slack [17:37] and look at member requests and try to [17:39] process as much of it as it can by [17:41] itself. I have it automatically process [17:44] refunds and disputes, figure out email [17:46] mismatches, and so on. And here's a [17:49] quick example of something that I [17:51] proactively handle for me. So, someone [17:53] sent me an email asking for a lifetime [17:55] plan, and right now I only have annual [17:57] plans. So, it proactively drafted this [17:59] email saying that, "Hey, we don't have [18:01] lifetime plans, but I'll put it under [18:03] consideration." And it asked me if I [18:05] want to send this reply or not. And this [18:08] comes down to the crux of why I really [18:10] like Rockbot. It's a really polished [18:11] piece of software. But because it's [18:14] based on bots and personas instead of [18:16] chat threads, it really does feel like [18:18] hiring employees to work for you. And [18:21] each employee takes care of a specific [18:22] job like customer support for behind the [18:25] growth, YouTube research and prep for [18:27] YouTube producer, helping my wife buy [18:30] stuff and book family stuff with loving [18:33] husband, weekend planner, planning trips [18:35] and stuff, and keeping my health on [18:38] track as health coach. So these bots [18:41] instead of random chat threads just feel [18:43] like organized employees or helpers to [18:46] make my life easier. All right, so let's [18:48] do a quick recap. Here are the use cases [18:51] that we covered. Number one, tagging bot [18:54] on X triggers it to work inside Grockbot [18:57] to do some analysis for you from any [19:00] kind of X tweet or X reply. Number two, [19:03] now that Grockbot has free access to X [19:05] data, you can do all kinds of awesome [19:08] things like sentiment analysis for your [19:10] brand or product, analysis of your own [19:12] tweets to figure out which ones go viral [19:14] and why, getting YouTube ideas by [19:16] combining X data with YouTube data, and [19:19] potentially down the line also analyzing [19:22] and responding to your DMs. Number [19:24] three, Grockbot now has it own email, so [19:27] you can copy it in an email thread for [19:29] it to schedule a meeting for you or do [19:31] some research. And as I mentioned, what [19:33] I find really useful is having it pull [19:37] data from multiple sources and send me a [19:39] bunch of email reports that are just [19:41] super insightful and actionable. Number [19:43] four, you can also add multiple Grock [19:45] bots to the same chat thread, which is a [19:47] unique feature of this product and have [19:49] them banter and figure out things like [19:52] for example consolidating how many [19:53] YouTube research emails I get. And [19:56] finally, as I mentioned, Grockbot is now [19:58] powered by Opus by default. You know, [20:01] I've been wondering when Anthropic will [20:03] launch its own Grockbot competitor, but [20:05] for now, Grockbot is the best place to [20:08] use Anthropics models. And last but not [20:11] least, we walk through a real use case [20:14] where I built a bot to provide customer [20:17] support for my paid community. And you [20:19] can do something very similar as well. [20:22] Okay, so that's it. I always want to [20:23] give you guys a real deal. And for now, [20:25] Grockbot legitimately is my favorite [20:27] personal agent. I think it's more [20:30] reliable and cleaner than chat shifty [20:32] dots and it's also more powerful than [20:34] Muse with Opus. So definitely give [20:37] Grockbot a try. And as always, please [20:40] like and subscribe for more practical [20:42] nononsense tutorials like this. I'll see [20:44] you soon.