Social Done. Search Next.
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 experiments with a multi-agent system and an automated, multi-platform social-content pipeline. Luqman discusses a possible shift from fixed loops and assigned tasks toward graphs and goal-driven agents with roles, contracts, memory, and hierarchy. He presents these as evolving ideas and experiments, not proven capabilities: agent communication cost, usage limits, stop controls, review, and reliability remain open problems. The practical work includes plans to generate video, text, voice-over, and captions, run the pipeline overnight with scheduled checks, and move gradually from human review toward automation while keeping personal posts under the account owner’s review. The stream does not establish that this end-to-end run was completed successfully. Audience questions broaden the discussion to connecting WhatsApp with AI tools, selecting models by total task time, retries, and usage rather than names alone, and protecting human judgment from model flattery or excessive autonomy. Other substantial topics include validating user problems, pricing around measurable usage costs rather than outcomes that cannot be guaranteed, the difficulty of acquiring traffic, and learning marketing, programming, and practical skills in the AI era. Additional discussions cover app-publishing and incorporation considerations in general, non-definitive terms; fair and secure agent competitions; visual summaries for long texts and streams; a proposed experiment with AI-generated musical variations; and lighter sports and workspace topics. The clearest conclusion is a preference for hands-on learning and building with human oversight and explicit operating limits. GitHub releases, autonomous operation, the virtual-team architecture, and several proposed tests and integrations remained plans or unresolved questions rather than confirmed results.
82 moments from the stream
- 2:21:15
Learning Marketing Fundamentals with AI
The speaker suggests that strong self-learners can use AI to understand marketing fundamentals and follow sources that provide broad guidance, while noting that he has no specific specialist to recommend here.
- 2:24:51
An Opinion on Alex Hormozi’s Content
The speaker characterizes Alex Hormozi’s content as broad rather than focused marketing instruction and feels it has partly shifted toward motivational material; this is the speaker’s personal assessment.
- 2:25:26
Making a Building Competition Fair and Safe
The speaker and audience brainstorm coding or business competitions, including building a product or a chess game and selecting the best result. He stresses that the reward must be fair to avoid exploitation and is wary of directing participants toward a live site because of security risks; the final competition format remains undecided.
- 2:29:27
A Reported Impression of Impeccable for Design Improvement
The speaker reports that someone tried the Impeccable skill and was surprised by the degree of design improvement. This is a reported impression rather than a documented test, and the following recommendation in the captions is unclear.
- 2:31:17
An AI-Agent Chess Competition
A proposal for each participant to design a chess-specialist agent and have the agents compete. The discussion highlights chess’s near-absence of chance, its long history with AI, and the qualification that the game has not been fully solved.
- 2:31:47
Giving a Coding Agent Discretion to Use Skills
A recommendation to try a skill with Codex, alongside an approach that lets the coding agent decide when the skill is useful instead of requiring it for every task.
- 2:36:42
Preventing Cheating in Agent Competitions
The discussion notes that entrants could cheat by using an existing or purpose-built chess agent, so the competition would need an implementation plan and strict rules. The idea remained exploratory, with openness to finding a better challenge.
- 2:38:04
A Graph for Tracking an Automation System
A brief explanation of moving from a graph-based bot to a graph representing the automation system itself. Presets translate what happened in the workflow, with links to files and agents; the system is still being developed and needs refinement.
- 2:41:17
Do Not Fully Outsource Your Thinking to AI
A caution that sharing some personal context can help an assistant, but AI should not completely replace the user’s own thinking and judgment.
- 2:42:59
Preserving Chat Context While Starting a New Session
A practical workflow: preserve the current conversation for later continuation while opening a separate session for a new task.
- 2:44:02
A Brief Opinion on Perplexity for Medical Research
A brief personal preference for using Perplexity in medical research, without a documented comparative test.
- 2:48:10
A Personal Comparison of Computer-Use Features
The speaker gives a personal opinion that GPT’s computer-use capability is much better than Claude’s; the captions do not describe a systematic test.
- 2:49:32
The Human as AI Orchestrator, Not a Recipient of AI Decisions
The speaker compares the human role to that of a manager rather than an executing employee: directing AI tools and making decisions. In his view, risk arises when a tool decides on its own or adds features without human direction.
- 2:51:41
Learning Agent Systems by Building Them
The speaker says that building working agents over a short period taught him more than studying the subject only in theory, and that the experience revealed the value of flexibility and control in improving workflows.
- 2:53:37
Monitoring Automated Workflows with Scheduled Checks
The speaker describes testing an overnight content-production workflow with a scheduled job that periodically checks the system for problems and attempts to address them. He presents it as an experiment in progress, not a proven success.
- 2:55:48
Does Human Advantage Shrink as Models Improve?
A participant argues for preserving room for human reasoning when managing agents and solving problems. The speaker replies with his view that outperforming models may become harder as they improve, citing a claim about many agents collaborating on a longstanding mathematics problem; the model details and event are unclear in the captions and are not presented here as verified facts.
- 3:00:01
Challenges to an AI Mathematics Claim
Audience members raise questions about whether the reported solution story is accurate and where its ideas came from, including allegations that an individual’s earlier work may have been used without proper credit. The speaker offers his interpretation that the system may have drawn on extensive available material, but does not establish what happened; the claims remain disputed and uncertain.
- 3:03:55
Turning an Organizational Explanation into a Visual Diagram
He asked for a long textual explanation of a department and its hierarchy to be converted into an easier-to-follow graph, explaining that visual relationships help prevent details from becoming tangled. He compared this with his earlier use of diagramming tools to organize ideas, work phases, and brainstorming, and suggested that models may streamline the process through a prompt, without demonstrating a confirmed result in these units.
- 3:07:50
Balancing Deep Reading with Live-Stream Pace
He explained that he enjoys reading but does not want to spend a long stretch reading an entire text during a live stream when that could slow the presentation or bore viewers. He therefore prefers condensing lengthy material into a quickly understandable visual graph and said he generally favors books with practical value.
- 3:11:34
Reading and Visual Summaries of Long Texts
The speaker values reading and speaks positively of The Forty Rules of Love, while noting that automated summaries are not always good. He also demonstrates condensing lengthy written material into a visual graph when reading it all is impractical.
Got a question about something in this stream?
Ask in the group