New Project? New CC?
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 vision for Mission Control as a unified layer for monitoring agents and sub-agents, routing work by goals and available capacity, and moving from small tasks toward longer, more autonomous builds. Luqman discusses preparing PRDs, ERDs, system design, infrastructure, and access in advance while keeping humans in the loop and adding code, security, and Playwright checks before merges; however, the design, integrations, and full automation remained exploratory, with no confirmed completion. Audience questions covered choosing tools such as Codex and Claude, context and usage management, learning Python versus AI engineering, and reverse engineering with its legal risks. Practical discussions also addressed product distribution and audience reach, email-deliverability testing, AEO for Shopify stores, interface improvement, and repository knowledge. A recurring attributed view is that focus, quality, distribution, and reviewing AI-produced work matter more than maximizing project or content volume, while several technical, legal, and experimental outcomes remained unresolved.
103 moments from the stream
- 2:24:25
Find Project Ideas in Everyday Needs
When asked for portfolio project ideas, the advice was to begin with a personal problem or daily task, then look at what people nearby do and build something that helps them. Practical project ideas can emerge step by step from real needs.
- 2:31:55
A Proposal to Release a Tool Under the MIT License
An audience member asked whether a discussed tool could use the MIT License so it could be combined with another tool. The speaker agreed to do so if it was published. The projects’ identities and spellings are uncertain from the captions.
- 2:33:08
Questions About Agent Control and Plan Continuity
The speaker questioned why the agent could not take over the browser task directly and compared that limitation with what he expected from Codex. After forking the conversation, he also asked where the original plan and its progress had gone; the captions do not provide a conclusive answer about these capabilities.
- 2:34:50
Questions About Email Address Warm-Up
The speaker reviewed an email warm-up plan and asked why it proposed one Gmail and one Outlook address, and whether that choice was specifically for warm-up. He also raised, without a clear resolution in this excerpt, whether addresses voluntarily supplied by participants could help.
- 2:36:18
Replacing an Exposed Private Key
After noticing that a private key had been exposed during setup, the discussed response was to delete it, generate a replacement, and download the new PEM file. No key or credential fragment is included in the index.
- 2:38:04
Using Claude and Codex Together
Asked about choosing between Claude and Codex, the speaker said he works with both to roughly the same extent. He began to explain why he was not using Claude at that moment, but the excerpt ends before the reason is given.
- 2:38:30
A Personal Comparison of AI Tools
The speaker compares his experience with several AI tools, personally leaning toward Claude for quality despite not seeing a large difference, and prefers letting the tool that originated an idea continue developing it.
- 2:46:57
From Step-by-Step Work to Autonomous Project Building
The discussion shifts from handling one step at a time to giving the full project outline to another session to drive toward completion. This first requires setting up infrastructure, accounts, the stack, and APIs before deployment and a more autonomous build experiment, with time savings as the main expected benefit and usage limits as a potential constraint.
- 2:50:51
Funding AI Usage as a Competitive Advantage
The speaker argues that funding intensive use of tools such as Codex can give a product a competitive edge, and that funding supports production itself, not only marketing.
- 2:56:12
Preparing Sub-Agents for Autonomous Building
The speaker explains a goal of anticipating blockers and seeing the full roadmap so sub-agents can keep building and resolve obstacles more independently.
- 3:00:05
A Platform as a Continuously Evolving Product
The speaker describes a platform as something that never reaches a final endpoint but keeps evolving, guided by a core vision for continuous and increasingly autonomous development.
- 3:00:27
Concept for an Arabic E-Learning Marketplace
The presented documents describe an Arabic-first e-learning marketplace for the MENA region, with student, instructor, and admin roles, course-state workflows, and phased scope covering an MVP and roadmap.
- 3:03:27
Learning While Building Software with AI
Asked whether to build manually or with AI, the speaker recommends building alongside AI and using PRDs, system design, and ERDs as part of the learning process. In his view, this approach can greatly reduce development time while still building experience.
- 3:05:11
Comparing Claude, Codex, and Cursor for coding
The speaker advises someone with a software-engineering background to evaluate Claude and Codex according to budget and actual usage. As a personal opinion based partly on what he has heard, he describes Cursor as a strong harness for Claude and recommends monitoring token consumption or usage limits before choosing a higher subscription tier.
- 3:10:04
Trialing two tools before committing to one
The speaker suggests trying Codex and Claude for an initial period, comparing their efficiency and fit for the work, and then concentrating later spending on the preferred tool. He also notes that cheaper models exist, while expressing the opinion that they may not provide the same level of performance.
- 3:11:55
Can Skills Reduce Token Usage?
The speaker says skills may save some tokens but questions whether they always offer the best quality. He suggests using lighter models for routine tasks and stronger models for demanding work, while stressing that this is a personal and inconclusive assessment.
- 3:19:23
Automating Lead Research and Outreach
The discussion describes a workflow that researches high-intent leads and sends outreach emails. The speaker compares handling research and outreach within the current workflow with using an external service he considers more capable, without claiming verified results.
- 3:22:33
Preparing Cloud Infrastructure for Autonomous Agents
After noting that the cloud environment is ready for development, the speaker asks for the day's files to be submitted as a pull request and for setup work to continue toward more autonomous operation.
- 3:24:57
Email Seed Testing with Smartlead
The discussion distinguishes testing from warm-up because Smartlead handles warm-up internally. It gives an estimate that about 10 seed addresses can provide a usable reading and 15–20 a cleaner one, while emphasizing that provider diversity matters more than count alone.
- 3:31:48
Building a Project Command Center
Luqman recommends creating a command center or “Mission Control” for each project: a map of its milestones with a task board for tracking stages such as the build and pilot. He said he used this approach while building Burnish in March or April 2026, and that a more complex new project requires a different command center tailored to its particular vision. He also said he might publish the command center as open source on GitHub so the community could contribute.
Got a question about something in this stream?
Ask in the group