Back to Warp.dev? 2x Claude + Codex
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 indexed portion of the recording centers on coding-agent orchestration. Luqman describes using Codex to supervise Claude sessions through computer use and scheduled checks, allocate tasks, hand off context, and review work. Audience discussions raise workflow visibility, lost details, and trust in reviews, particularly for scientific code; reliability and token efficiency remain unresolved. He explores a goal-driven workspace combining agent teams, memory, roadmaps, and an interface that adapts to each project. Practical work includes reviewing system architecture, extending plans beyond MVP, and requesting landing-page and hoopoe-animation revisions. The concept and revisions remain in development, alongside discussion of open sourcing; completed implementation and fixes for computer-performance problems are not established. Audience questions cover getting started with AI, programming fundamentals, hardware requirements, security, and developing skills. Other discussions examine task cost versus token pricing and model speed without a measured comparison. On product strategy, Luqman emphasizes lasting usefulness, discovering problems for a focused audience, and planning marketing and distribution. He argues that easier building does not guarantee sales and that marketing outcomes remain uncertain. Secondary topics include agents that read and reply to Facebook and TikTok comments while identifying themselves as AI. Luqman reports workflow problems, local slowness, and agents stopping when the application or computer closes, without documenting fixes. Participants also discuss voluntary creator support through tips and paid messages, affiliate marketing, sponsorships, personal branding, and professional visibility through building in public; the question of how personal branding relates to technical competence remains unresolved. Discussion also covers AI-company IPOs and profit-and-loss disclosure. At recording time, Luqman offers unverified forecasts about IPO timing and funding improving pricing and efficiency, while audience contributions suggest funding could support data centers and research. He also discusses his trust in expert forecasts and a warning attributed to Geoffrey Hinton about superhuman-AI risks, presented as a risk forecast rather than an established outcome.
160 moments from the stream
- 2:20:41
Why vibe coding should not be an insult
The speaker rejects treating the vibe coder label as evidence of ignorance or unwillingness to learn. He values building products that achieve market results, while emphasizing respect for developers and his view that AI makes programming easier to learn.
- 2:23:15
Is a MacBook Air enough to start building with AI?
Answering an audience question, the speaker says a MacBook Air is sufficient for learning and getting started, provided models are not run locally, and notes that stronger hardware can improve the experience.
- 2:23:45
An initial understanding of AI-assisted reverse engineering
The speaker begins discussing reverse engineering and tentatively suggests that AI reads binary code, explicitly acknowledging limited understanding. The passage ends before he explains the resulting operation, and the tool name is unclear.
- 2:25:28
Motion graphics with AI
Asked whether to leave motion graphics, Luqman recommends continuing in the field and using AI to improve speed and quality. He believes demand remains strong.
- 2:26:36
Entering AI engineering from mechatronics
Luqman argues that a strong portfolio and practical experience can offer a route into AI engineering without a computer science degree, while expressing his view that major companies require a specialized academic background. He also connects mechatronics with robotics.
- 2:28:56
Reply drafts and less manual prompting
Luqman describes accepting suggested reply drafts with Tab in Claude and Codex, saying Claude introduced them first according to his observations at the time. He recommends reading them initially and says they sometimes improve on his own wording, within a workflow he describes as increasingly autonomous.
- 2:32:05
Chat context and choosing Claude Chat or Claude Code
An audience question raises context continuity across Claude chats and between chat and Claude Code. Luqman expresses a preference for Claude Code for work and projects, including video work; the passage does not provide a technical solution for sharing context.
- 2:32:53
Starting with AI before mastering the terminology
Luqman recommends starting with a conversation about one's background and goals, asking the tool to pose clarifying questions and suggest practical steps. He favors a paid service according to budget and suggests Claude or Codex in this discussion. A later remark suggests sometimes treating AI as a manager or senior colleague.
- 2:37:48
Is ComfyUI open source?
ComfyUI comes up briefly: Luqman recognizes the name and asks whether it is open source. These pages contain no confirmed answer or usage explanation.
- 2:39:23
Does using AI require a powerful computer?
Acknowledging that hardware is not his specialty, Luqman says using AI tools without running local language models does not require a powerful computer. He provides no specific hardware requirements or benchmark results.
- 2:39:59
Showing the process and setbacks of building tools
Luqman explains that the stream documents using AI to develop skills, build tools, and try to sell them, including setbacks without assuming success. He argues that short success videos can make the process appear easy.
- 2:41:28
Securing a website built with Codex
A viewer asks about vulnerabilities in a website built with Codex. Luqman says he cannot provide a specialist security answer and suggests consulting someone with strong security knowledge and also asking AI; he provides no specific remediation steps.
- 2:42:19
Communicating your idea before choosing Claude skills
Asked for the most important Claude skill, Luqman does not name one. He argues that beginners should first learn to communicate their ideas to AI, with additional skills coming later to refine the work.
- 2:43:12
Vibe coding revenue depends on reaching a market
Luqman says products built through vibe coding can generate revenue, citing a commercial example whose identity is unclear in the captions. He stresses that coding alone does not produce profits: entering a market, selling, and reaching an audience are essential.
- 2:44:37
An audience suggestion to apply security skills after every change
An audience comment broadly claims that security skills can protect against vulnerabilities and recommends applying them after every change; Luqman agrees. This passage provides no assessment or evidence establishing comprehensive protection.
- 2:44:48
From planning documents to a trackable roadmap
Responding to a system-design question, Luqman describes his personal planning framework through requirements documents, a roadmap, and a manifesto. He discusses bringing milestones and tasks into one clear place and acknowledges that having an ambitious vision does not yet mean knowing how to implement or explain it.
- 2:47:10
Starting vibe coding within your existing field
Luqman argues that starting to build with AI does not require already being a programmer, while recognizing the value of programming fundamentals and learning through practice. He agrees with an audience suggestion to apply vibe coding within an existing field of expertise rather than starting without experience in another domain.
- 2:50:10
Forking a conversation to organize a project roadmap
Luqman describes one session overseeing four others, then forks a conversation specifically to organize a roadmap. He requests a project section in an existing tracking application showing tasks, milestones, and missions to make progress easier to follow. The passage captures the request, not a confirmed completed implementation.
- 2:52:50
Learning AI tools and the limits of relying on livestreams
A viewer asks whether livestreams can teach AI tools without a technical education. Luqman says the streams are not a course and recommends starting a conversation with an AI tool about one's background and ambitions. He also describes a planned keyword-searchable stream library whose final placement was still undecided at the time.
- 3:01:13
A project roadmap for tracking progress and reviewing plans
The speaker discusses a roadmap organized into milestones, missions, and baseline tasks, with work underway on the first milestone. He acknowledges remaining uncertainty about the work and the need to review the plan, agreeing that a roadmap helps the audience follow progress, especially if the project becomes open source.
Got a question about something in this stream?
Ask in the group