One AI Department Done. 3 to Go.
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 a work-in-progress automation system that organizes AI agents like connected departments for research, coordination, content production, publishing, and analytics. Luqman explains a possible progression from scheduled tasks and role-specific instructions toward shared goals and indicators, with research handed to agents that adapt posts, articles, and short videos for each platform. Practical work includes reviewing the post-stream pipeline, trimming silence and adjusting clips, and preparing operations across LinkedIn, X, Facebook, and Instagram while retaining human review initially. Speed, quality, resource use, queues, measurement, and agent autonomy still require testing and refinement; the stream does not establish a successful end-to-end system. It also discusses distribution and building in public, comparing direct outreach with SEO, social content, and advertising, alongside the speaker’s view that understanding an audience and solving a real problem matter more than selling a generic chatbot. Audience questions broaden the discussion to changing developer roles, learning and working with AI, game development, video-editing tools, background or local agent execution, community platforms, tools, and subscriptions. Other cautious, unresolved topics include AI jobs and pricing, proposed Egyptian data centers, model and hardware performance, YouTube Shorts growth, and ideas for competitions and products; these are generally presented as opinions, proposals, or open questions rather than verified findings.
171 moments from the stream
- 5:22:32
AI Tools for Faster Video Editing and Stream Repurposing
The speaker recommends exploring AI editing tools and cites an audience comment saying that DaVinci Resolve has good MCP support; he then gives his opinion that tools of this kind may improve speed and quality. He also describes an automated stream-repurposing workflow that selects notable moments, removes silence, adds captions, and formats vertical clips to a configured layout, while qualifying that the output may not match skilled human editing.
- 5:24:42
Breaking Down Complex SOPs with Codex or Claude
The advice is to try Codex or Claude with a capable model to analyze a complex SOP, clean up the instructions when needed, and work through it one step at a time while recording what happens. The passage ends during a question about which parts of the SOP could be improved, so the final recommendation is incomplete.
- 5:25:39
Testing What AI Can Handle in a Workflow
The speaker suggested showing a task or workflow to an AI tool and asking which parts it can perform. He expected this to simplify and accelerate the work, while explicitly noting that he had not seen the resulting output and could not vouch for its quality.
- 5:28:02
AI Tools for Clipping Highlights from Long Streams
The speaker discussed why gaming clips cannot be selected from speech alone: important action may happen during silence, while speech may occur during a dead or inactive scene. He described gamer-oriented tools that analyze frames, speech, and action to propose moments, try not to cut sentences, and allow extending a clip before and after the selected point, presenting them as a possible way to reduce manual review of multi-hour streams.
- 5:31:53
Building a Music Arrangement Around a Recorded Melody in Suno
The speaker suggested recording a short tune, bringing it into Suno’s studio, and then adding another audio track or a guitar-melody layer over a background element such as a pad. The idea was presented as a practical way to experiment with arranging layers around recorded musical material.
- 5:36:12
Reviewing Content Agents Before Granting Autonomy
The speaker describes reviewing content agents’ output and requesting revisions when needed, with plans to reduce manual oversight gradually after their work proves reliable, especially for personal accounts.
- 5:39:28
Options for Running Agents in the Background or on Dedicated Hardware
Answering an audience question, the speaker compares a virtual environment that can continue after his device is closed with hosting agents on dedicated hardware that remains available. He stresses that the final setup is undecided and that the priority is to activate and understand the system before expanding continuous operation.
- 5:41:34
Automating Screen Capture as Video or GIF
The discussion calls for fully automating the process and mentions screen recording, with a GIF as an alternative to video. The implementation details remain incomplete on this page.
- 5:44:49
Preventing Recurring Errors and Working with Agents Inside a Project
The segment emphasizes ensuring that an error does not return after being fixed, while noting that an agent may struggle to anticipate unknown future cases. It also says the agents operate within the project files and can be addressed through Codex or Claude, though the explanation continues beyond this page.
- 5:45:40
Organizing an AI-Agent Team with Roles and System Prompts
The speaker describes treating each conversation as an agent with a defined responsibility, task, and operating contract in its system prompt, including a research-to-review handoff that filters duplication and distributes results. He distinguishes this current design from an unfinished idea of giving agents a shared goal instead of fixed detailed instructions.
- 5:48:03
From Prompt Engineering to Loop and Graph Engineering
Responding to a suggestion to use a prompt-writing agent, the speaker argues that an elaborate prompt is not always the central need; clearly stating the desired outcome and designing loops matter more in his experience. He attributes to a recent Codex article a progression from prompt engineering in 2023 to loop engineering and then graph engineering.
- 5:51:25
What to Learn for AI-Enabled Work
The speaker advises the questioner to continue learning Python without rushing, while also developing market and sales knowledge. He argues that a clear personal direction, followed by giving an AI model appropriate context about capabilities and ambitions so it can suggest a path, may be more useful than pursuing an isolated skill without a goal.
- 5:53:02
An Automated Agent for a Daily AI News Brief
The speaker proposes an agent that periodically gathers AI news from several platforms and sources, summarizes recent developments in a report or webpage, and delivers the brief through messaging or to a computer. It is presented as a possible practical agent use, not as a system demonstrated as completed here.
- 5:57:37
Balancing Quality and Cost in AI Video Generation
The speaker discusses the tradeoff between quantity and quality in AI video generation. In his opinion, advanced paid tools may deliver more value and quality than open-source alternatives, while open source can still reduce cost; he also describes an internal experiment that still needs speed and prompt improvements and mentions Remotion as a possible option.
- 6:03:09
Livestreaming a Product-Building Journey Transparently
The speaker says the channel aims to livestream the full working journey transparently rather than only publish short tool comparisons. That journey extends from coding and building command-line tools to market entry, marketing, and traffic acquisition, showing either success or failure; he expects regular Arabic livestreaming about AI and coding in this format may become more common.
- 6:06:33
Usage Limits in AI Tool Subscriptions
The speaker criticizes an AI tool’s usage model after a large portion of its allowance was consumed quickly, saying the constraint made it unsuitable for his workflow. The tool’s name is not clear enough in the captions.
- 6:07:38
Canva MCP Helps with Images, Not the Video Problem
The discussion suggests Canva’s professionally designed post templates, but the speaker says images are not his bottleneck. His main problem is video, which he does not expect Canva to solve in this case.
- 6:09:23
Telegram or Discord for an Arab Community?
The speaker says he initially favored Discord, but considers Telegram much more familiar in the Arab world. He views Discord as a niche option that could be reconsidered if the community grows substantially.
- 6:09:56
Choosing an AI Tool for UI/UX Work
As a personal recommendation, the speaker favors an AI tool whose name is unclear in the captions for interface work and suggests describing the task directly to it. He adds that a different tool may suit a specific website motion effect, though the other tool names are not clear enough to identify.
- 6:11:43
Checking Public and Private Visibility
During the discussion, the speaker discovers that material expected to be non-public is publicly visible, so he decides to make it private while leaving another item public. The exchange highlights checking each item’s visibility before release.
Got a question about something in this stream?
Ask in the group