Building Grok Bot | 5 Executives, 16 Employees
Want to back the stream? The «داعم» membership on YouTube is EGP 60 a month: badges, emoji, and I answer your questions first on stream. Become a supporter
The gist
The stream centers on an unfinished experiment to build a Grok Bot agent workforce with a clear hierarchy, role charters, cloud environments, permissions, handoffs, recurring meetings, contribution metrics, and human review before publishing. Luqman prioritizes marketing and organic traffic over further feature work, discussing programmatic SEO, stream repurposing, social channels, and deliverable email outreach while warning about random content changes, sensitive integrations, and broad agent access. Audience questions prompt comparisons among Codex, Claude, and Grok Bot; he stresses that selection should follow the task, efficiency, and cost, and that his views are early impressions rather than systematic findings. Other discussions cover engaged communities, freelancing and practical AI learning, possible labor-market effects, and AI-assisted music. The impact of the marketing agents, the value of the organizational design, usage-limit consumption, and several automation workflows remain unresolved; smoke testing was not completed and the overall system was not shown to be successful.
139 moments from the stream
- 3:54:29
Open Question: Is Grok Bot Worth Switching From Claude?
An audience member was asked for a general assessment of Grok Bot and whether it justified moving from Claude, but this page contains no answer or comparative conclusion.
- 3:58:15
Evaluating Agent Environments and Cloud Computers
The speakers discuss evaluating agent tools by cost, freedom to choose a model and environment, and the availability of a cloud computer for each agent. They clarify that some options have not yet been tested and suggest adopting tools when they fit the current stage and practical need.
- 4:03:11
An Agent for Turning Streams into Short Clips
The discussion proposes training an agent to feed stream recordings into a clipping tool and produce short-form videos, then using those clips as part of a broader audience-reach strategy across channels such as Telegram and TikTok. It is presented as a possible workflow, not a completed or tested result.
- 4:07:06
Automating email outreach while protecting domain reputation
Explains using a dedicated sending platform to schedule outreach rather than sending directly through Gmail, with a protected sending domain to reduce damage to domain reputation and spam placement. The exact implementation remained dependent on the intended workflow.
- 4:08:46
Designing a clear, non-redundant dashboard banner
A product-interface review identifies the visibility banner as the strongest option, while the activity banner repeats information found in other tabs and uses unclear labels. The direction is to make the banner concise, useful, and functional.
- 4:15:13
A review and correction loop before publishing content
Discusses a workflow in which drafted content is reviewed and returned for correction when a statistical error is found before publication, acknowledging that immediate publication without this step was unsuitable.
- 4:16:12
Stress-testing the scaling logic of programmatic SEO
Challenges batching publication and imposing a low article ceiling, describing programmatic SEO as broad production across many keywords. This is qualified by the observation that a newer domain may be indexed slowly, supporting a gradual ramp, followed by a request to list the dependencies needed before article writing begins.
- 4:20:15
Using a browser agent to automate Gmail tasks
Proposes delegating repetitive Gmail steps to a worker or agent operating through Chrome, with task recording available when an instruction is ambiguous. It is framed as an experiment, with the qualification that automating everything is difficult and some small issues remain unresolved.
- 4:28:21
Measuring the Workflow and Assigning Ownership
Questions were raised about how to measure the full workflow, including domain authority, and who should be responsible for tracking the measurements.
- 4:31:06
Choosing Stream Language, Platforms, and a Planning Method
After initially considering both English and Arabic, he preferred making the stream more Arabic. Luqman says the streams are not aimed at virality or profit and considers matching language to platform: English content on X and LinkedIn, while remaining uncertain about Twitch and Kick because of their audiences. When the options become tangled, he returns to pen and paper to organize his thinking.
- 4:32:32
Moving Forward with a Multi-Role Team
The discussion highlights the difficulty of progressing across many tasks and asks whether the work should proceed one employee at a time.
- 4:35:49
Trying Grok Bot While Continuing with Codex
Luqman considers setting aside periodic time to try Grok Bot while continuing to use Codex, noting that he is waiting for certain unspecified Codex capabilities.
- 4:36:45
Codex for Automation and Coding
The speaker considers Codex excellent for automation and capable of handling a great deal of work, while acknowledging that it can sometimes be slow. He also anticipates a future tool or release whose name is unclear, but remains uncertain about its resource usage and practical usability.
- 4:38:27
AI Tool Usage Limits and the Superintelligence Race
The speaker says some Claude solutions can feel over-engineered, then recalls user complaints about changed usage limits in late March 2026, while stressing that what happened inside the company is unknown. He presents his view that major AI companies are racing toward superintelligence because the first to arrive could capture substantial market share, naming Grok, Claude, OpenAI, and Google as participants; this is personal analysis, not a confirmed account.
- 4:42:30
Stream Music and Copyright Risk
Answering a question about the music, the speaker says it could expose his stream to a copyright ban. He describes it as ambient music based on Pink Floyd remixes and suggests searching YouTube for similar material.
- 4:46:16
Audience Engagement Quality Over Subscriber Count
The speaker says he prefers a smaller, strongly engaged audience over a large subscriber base with only limited participation, placing the quality of the audience relationship above the channel’s headline size.
- 4:46:46
Why Audience Growth Feels Harder on YouTube
Based on his experience publishing across several platforms, Luqman says gaining subscribers on YouTube feels substantially harder than gaining followers on TikTok, even when content is posted across the platforms in parallel.
- 4:52:55
Raising Awareness of Practical AI Use
After hearing that a viewer benefited from his content when using AI tools at work, Luqman explains that the channel aims to raise awareness of AI capabilities and practical applications.
- 4:55:08
Tracking the Impact of Marketing Agents
Luqman says he is nearing completion of a group of marketing agents and plans to observe their effect on traffic over the following weeks; the outcome remains unconfirmed.
- 5:01:09
Securing Income While Gradually Building a Freelance Path
The speaker advises someone needing near-term income to first seek stable work, then develop and market a side skill or service through freelance platforms such as Upwork. He stresses that freelancing is not instant income because a strong portfolio takes time, while noting that clients still seek people to handle web projects despite the availability of AI tools.
Got a question about something in this stream?
Ask in the group