Meen el hodhod dah?
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
A practical stream centered on an unfinished “Mission Control” tool for visually supervising coding-agent swarms, exposing their status and consumption, and detecting stuck loops, alongside live reviews of the interface, bird characters, and animation. The discussion broadened to goal-based roadmaps, separating software-building tools from company-operations systems, and balancing custom agent harnesses against added complexity and waste. Luqman answered audience questions about Codex, Claude, AI tools, learning, fundamentals, and hosting, repeatedly framing many comparisons as personal experience rather than conclusive tests. He also argued that distribution and demand validation can be harder than technical construction, and discussed a possible open-source GitHub release with community contributions or a competition whose prize and judging details remained undecided. The interface, motion, roles, naming, and release plan were still incomplete, and neither real token savings nor practical success had yet been established.
141 moments from the stream
- 3:10:08
Agent Harness Engineering as a Double-Edged Tool
The speaker describes agent harness engineering as adding sub-agents and control layers over the default workflows of tools such as Codex and Claude, enabling a system to read, act, and modify. He says the area is gaining attention but cautions that added control can disrupt the workflow when poorly designed.
- 3:11:26
Choosing an Accessible Competition Prize
The speaker notes that a game-related prize may require compatible hardware and connectivity, making it neither neutral nor useful to every winner, and asks for suggestions for genuinely attractive prizes.
- 3:16:33
Designing a Contest Prize for Broad Participation
A discussion of how to choose a prize that encourages broad contest participation: cash has wider appeal but may feel less creative, while hardware or subscriptions are more specific and may not interest everyone. Other ideas include stating the prize value upfront, splitting it across several placements, or letting the winner choose among prize types.
- 3:23:33
An Opinion on Choosing a Mac Studio
The speaker advises, as a personal opinion, against choosing a Mac Studio below an M4 Max configuration and says the M5 generation was expected during the month of recording; this is a historical expectation, not a current confirmation.
- 3:24:40
Why He Prefers Practical Work to Educational Content
The speaker explains that he does not want to intensify tutorials or sell courses he believes are less useful than observing practical work, framing this as a personal stance about honesty with the audience and himself.
- 3:25:12
Using Spare AI Capacity to Review Developers’ Projects
An audience member suggests directing spare AI-tool capacity toward reviewing participants’ projects and producing useful reports, and the speaker restates the idea to confirm his understanding.
- 3:30:34
A Visual Interface for Monitoring Coding Agents
The speaker demonstrates a visual view of a work session in which multiple coding agents connect to a central orchestrator, with states for active and completed agents and areas related to code, GitHub, and tests. He says the goal is to spot stalled work, questions requiring his intervention, and unsuitable resource use while turning text-heavy session data into clearer visual monitoring.
- 3:32:34
A Buy-Then-Resell Console Strategy
The speaker discusses an unverified estimate for buying a PlayStation 5 to play GTA 6 until its PC release, then reselling the console to reduce the eventual cost. The idea is attributed to a personal conversation and is not presented as confirmed financial advice.
- 3:33:08
An Incomplete Question About Whether a Tool Is Worth Subscribing To
An audience member asks whether the speaker recommends subscribing to a tool shown in an earlier video, but the tool’s identity is unclear and the page ends as the answer begins, so no actual recommendation is captured here.
- 3:33:23
Building a Project-Specific Command Center
While inspecting a stuck section that was consuming tokens, Luqman said he would defer the fix. He advised viewers interested in a command center to review the upcoming material, then build a version suited to their own project rather than copying his setup directly.
- 3:36:48
Separating Claude Code from Claude Chat
Luqman’s view was that project planning and implementation can be handled inside Claude Code, while Claude Chat is better suited to general personal conversations. He recommended treating them as separate workflows instead of trying to carry a chat project’s history and context into the coding environment.
- 3:38:09
When Is Higgsfield Worth Using?
Comparing Higgsfield with using model APIs directly, Luqman said the choice depends on the task and workload: a direct route may be enough for limited work, while Higgsfield may suit heavier, longer-term use, though he noted that it is not the cheapest option.
- 3:39:01
Learn Programming or Understand the Market First?
Luqman said the decision to learn programming depends on the person’s goals, background, and genuine interest in it. When forced to give general advice without that context, he leaned toward understanding the market and its direction first because the state of programming could change quickly within a year.
- 3:42:02
Education and Work in the AI Era
Asked what universities should teach, Luqman described his personal preference for entering the market and gaining work experience early, while stressing that he does not recommend leaving formal education for everyone. If choosing a field of study, he suggested areas around AI and related engineering, though some specific discipline names in the captions are unclear. He added that he regrets missing out on university.
- 3:43:30
The Value of Education and Fundamentals in the AI Era
The speaker argues that theoretical study matters in some fields but is not always the best path. After a participant says their curriculum had not been updated since 2007, the speaker encourages them to apply their software-engineering concepts and fundamentals when using AI tools, emphasizing conceptual understanding over any one prescribed learning path.
- 3:49:10
Sharing Failures Alongside Successes
The speaker says his goal in sharing his journey is to present it transparently, including its difficulty and failures, because success-only portrayals on social media can leave young people with the false impression that everyone else is succeeding.
- 3:51:56
Which Software-Engineering Fundamentals Does a Developer Need?
The speaker describes practical fundamentals including project planning, requirements documents, diagrams and system design, together with GitHub workflows such as pull requests, merging, worktrees and branches. He presents these as skills he found necessary to learn and is still developing.
- 3:54:49
Putting the Work Deck First in the Interface
The speaker criticized placing the work deck at the bottom and requiring scrolling to reach it, proposing that it appear at the top with clearer visual hierarchy.
- 3:55:28
Balancing Quality, Cost, and Efficiency Across AI Coding Tools
The speaker expressed dissatisfaction with a UI result produced by GPT and said he was beginning to notice a quality difference, while giving his opinion that Claude costs more and GPT is more efficient. The direction of the quality difference was not stated clearly in the captions.
- 3:56:50
A Semantic Color System for the Interface
The speaker was not committed to green and yellow specifically; the key requirement was that every interface color carry a clear meaning so states and information do not become confusing.
Got a question about something in this stream?
Ask in the group