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
- 3:17:37
Naming Agent Employees by Role
A proposed naming convention gives agent employees a first name plus a clear role or task, while executive roles receive fuller names and titles, making the team structure easier to understand.
- 3:20:11
Human Oversight for Generated Content Publishing
The discussion distinguishes directly produced content from public-facing posts that should be approved first. It suggests closer checks early on because image or video generation may be poor, followed by delegated review of routine content while monitoring broad SEO indicators.
- 3:22:48
SEO Risks of Random Content Changes
A warning is raised that an agent making random daily content changes could hurt rankings, followed by an open question about whether the risk comes from producing content unrelated to the site’s domain. The page does not provide a definitive answer.
- 3:24:07
Validating Agent Readiness and Handoffs
An approach is described for reviewing each agent employee individually, confirming access to the required tools and instructions, and checking handoffs across sequential roles such as prospect collection, assessment, outreach, and attempted conversion.
- 3:25:10
Page Stability, SEO Rankings, and Citations
The speaker argues that Google needs time to evaluate a stable version of a page and that changing it daily may cause ranking volatility, while GPT citations may shift to other pages. He also cautions against releasing many pages at once, duplicate or keyword-stuffed content, and allowing an agent to alter an already successful page without accounting for the ranking risk.
- 3:26:13
Simplifying a Bot Organization Hierarchy
The discussion proposes arranging bots in a clear company-style hierarchy to increase useful output with less supervision. It emphasizes keeping the structure understandable rather than overly complex and judging it by practical results rather than hype.
- 3:28:29
Using Voice Messages to Reduce Meeting Time
The speaker describes a general preference for concise voice messages, suggesting that they can handle some discussions more efficiently than long calls while clarifying that this is a communication preference rather than a refusal to meet.
- 3:31:39
Agent Coordination and Contribution Measurement
The speaker describes an experimental structure that assigns agents to responsibilities and departments, allows bots to communicate directly, and supports scheduled stand-up routines. He stresses that the approach is still being learned and evaluated, with the goal of measuring each agent’s contribution and return before deciding whether the system is worthwhile.
- 3:33:44
An Early View on an AI-Agent Platform
Asked about a suitable platform for building agents, the speaker expresses early enthusiasm for Grok Bot but says he is still at the beginning of the experiment and does not yet know whether it will work out. The statement is therefore an initial impression, not a final recommendation.
- 3:34:19
Unexpected Paste Behavior While Using Claude
The speaker describes an unresolved incident in which text appeared in Telegram when pasting while Claude was controlling the computer. He does not establish the cause and only raises device integration as a possibility, without reporting a test or confirmed conclusion.
- 3:36:56
Grok Bot Usage Limits for Email Tasks
The speaker explained that estimating how many emails could be sent before reaching the usage limit was not straightforward, and expressed concern that setup work itself consumed the limit quickly, without claiming a fixed calculation.
- 3:38:05
Defining an Agent Workflow Before Activation
The discussion raised practical questions about the agent’s capabilities, when it starts working, what triggers it, and who owns each step. It highlighted that the end-to-end workflow was not yet defined clearly enough.
- 3:41:25
Cleaning a Mailing List Before Sending
The speaker proposed validating a large mailing list with ZeroBounce first to remove addresses likely to bounce, then using a sending service such as Resend. Periodic batches were also suggested instead of processing the entire list at once.
- 3:43:32
An Opinion on the Different Roles of Claude and Codex
The speaker said Codex was more effective for his coding work, while Claude should not be viewed only as a coding tool because he considered it broader in reasoning and discussion. This was a personal opinion, not a systematic test result.
- 3:45:11
Opinion: Claude Is Less Agreeable in Personal Conversations
The speaker says Claude seems less inclined to flatter or simply agree in non-coding personal conversations, and notes that he likes this trait.
- 3:46:09
Email Sending Limits Depend on the Provider
An explanation that sending capacity is tied to the email provider, so campaign design must begin by identifying the sending platform and its actual limits.
- 3:47:17
Balancing Agent Autonomy with Read-Only Access
A discussion of whether read-only permissions are too conservative when the goal is capable, autonomous agents, framing safety versus broader access as an open design question.
- 3:48:46
Designing an End-to-End Programmatic SEO Workflow
An outline of a programmatic SEO system covering publishing cadence, domain-ranking and traffic measurement, topic selection, generation and verification, ROI measurement, and the need for a diagram showing what triggers and supervises each step.
- 3:50:07
Why Sending Campaigns Directly from Gmail Is Not Enough
An explanation that having an agent send messages one by one through Gmail would be slow, while larger-scale outreach needs suitable infrastructure, distribution, and deliverability practices to reduce spam placement.
- 3:54:04
Distraction as a Response to Feeling Stuck
A general personal observation that when someone feels stuck and cannot see a path forward, they may focus on improving side matters, turning activity into a way of avoiding the main next step.
Got a question about something in this stream?
Ask in the group