5 TASKS IN 7 HOURS
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 follows work on a Shopify app and a multi-agent “Command Center” workflow that divides exploration, implementation, and auditing while keeping the human responsible for setup and sensitive decisions. Practical work centered on connecting AI APIs, aligning environment variables across Railway and Supabase, configuring Sentry for production monitoring, and troubleshooting deployment, authentication, trial, approval, and cancellation flows. Some deployments and checks were reported complete, but issues including an HTTP 500 error and parts of the user flow remained unresolved, illustrating the limits of automation and the value of clear logs, staged audits, and CI/CD. Luqman also discussed choosing coding tools by task, retained context, and usage limits; his comparisons of Codex, Claude, and Kimi K2 were framed as limited workflow impressions rather than controlled benchmarks. Broader topics included designing defensible scores for stores’ visibility in AI answers, the unsettled GEO-versus-AEO terminology relative to SEO, and validating markets, pricing, distribution, and product economics before accelerating development. Research-oriented discussion extended to LLM interpretability and the lottery ticket hypothesis, as well as Luqman’s views on Chinese and US model ecosystems, reliance on NVIDIA chips, and the need for accessible Arabic models and local pricing. Audience discussions covered AI’s effect on developer roles, code comprehension and debugging, parallel milestones and dependencies, agent-rule context costs, productivity habits, and peer learning. This portion also included additional practical discussions about workstation and operating-system choices, stream latency, development practices before AI, and programmatic video creation with Remotion. Preparing Shopify submission media and polishing the Command Center for a possible GitHub release remained future plans, not completed outcomes.
127 moments from the stream
- 4:27:17
When an Automated Workflow Becomes Manual
The speaker reflects on the contrast between an automation-heavy day and another dominated by manual, step-by-step work. He also notes down ideas about requirements documents and CI/CD that may improve the workflow later, while several tool names in the exchange remain unclear.
- 4:30:04
A CI/CD Suggestion for a Flutter App
An audience participant suggests using Codemagic with a Flutter app and a GitHub Actions workflow where pushing to a release branch triggers deployment. This is presented as an audience suggestion, not as a confirmed setup or tested result in the current project.
- 4:30:49
Giving a Coding Agent More Precise Verification Instructions
While reviewing a verification task, the speaker asks the agent to explain more precisely why manual verification is required, why the agent cannot perform it, and what should be inspected when the deployed build only shows a limited home page with no navigation to billing. The response says the critical behavior occurs in the live environment and cannot be fully proven by unit tests, while the current branch contains only a placeholder page.
- 4:32:54
A Commit-to-Automatic-Deployment Workflow
The speaker describes his understanding of the current workflow: a command-center button commits and pushes changes to GitHub, followed by automatic deployment to Railway. The description reflects his current understanding of the flow and does not establish that a particular deployment succeeded.
- 4:35:25
Building a CI/CD Pipeline for Pre-Merge and Deployment Checks
The participants discuss automatically checking updates when code is pushed to a branch and identifying problems before merging to production. They note that setting up an organized pipeline can be difficult and time-consuming initially, but can later improve control over what reaches users and how it is deployed.
- 4:40:37
What Sentry Monitors and When It Operates
The discussion clarifies that Sentry monitors application errors during use and provides reports that help identify their causes; it is not itself a check for the deployment phase. This follows uncertainty about its relationship to the hosting platform and CI/CD, with participants noting its value when deployment logs are insufficient for diagnosis.
- 4:47:17
Reviewing an Early Visual Concept
An early design review favors a simpler logo, a smaller icon if retained, and a visual concept tied to the product’s core idea. The comments are preliminary preferences and iteration directions, not final design approval.
- 4:52:17
A Hotjar-Like User Session Recording Tool
A side recommendation mentions a free tool called “Clarity” for recording website or app user sessions, compared with Hotjar. The exact product identity remained uncertain because several tools shared that name.
- 4:54:33
Verifying a Backend Flow Without Confusing It with UI Polish
The discussion explains that the visible page is not being reviewed for polish; it is a trigger surface for exercising a backend flow and checking whether the app can launch. It also distinguishes a partial check from a milestone-wide audit, which had not yet been completed.
- 4:58:27
Separating Manual Setup Requirements from Deployment Failures
While examining a trial-flow failure, the discussion separates required manual setup from later deployment failures. A possible dependency-related cause is presented tentatively, so no single definitive root cause is claimed.
- 5:02:57
When to Develop Locally and When to Verify on Hosting
Responding to a question about finishing work locally before deployment, the speaker says routine development happens locally, while occasional hosted deployments are used to verify connections and environment-level API configuration. The deployment approach would become more conservative once the application reaches a market-ready release.
- 5:05:39
Using Multiple Agents to Diagnose a Root Cause
After the issue continued exhibiting the same behavior and a log was available, the proposed instruction was to deploy multiple agents to identify the root cause and fix the issue decisively.
- 5:13:30
Can a Large Monitor Improve Productivity?
During a discussion of a large monitor, the speaker says its value is practical rather than cosmetic and that, in his experience, it makes a real difference to productivity.
- 5:17:39
How Much Code Do Developers Understand With AI Tools?
A discussion of how much developers understand errors and code when working with AI. The speaker says he understands the broad structure and backend code better than programming-language details, while the discussion notes that modern error messages may identify the source and cause, suggest fixes, and connect to chat features inside many IDEs.
- 5:19:42
How Developers Solved Problems Before AI
A comparison between present-day debugging and earlier workflows, when developers searched for solutions and consulted Stack Overflow. The speaker wonders how complex software was built around 2010 and 2015 before modern AI tools, while recognizing that programming required learning a language.
- 5:22:48
Trying to Give a Coding Tool the Error Log and a Screenshot
During a practical attempt that is still failing, the speaker refers to the error log and plans to give the tool a screenshot for additional context. The passage does not establish that the issue was solved.
- 5:24:36
Programming Tools Before AI Assistants
The discussion recalls that a programming problem could once take a long time to resolve, then asks how developers managed large amounts of code and files when building applications before AI assistants. The speaker assumes other tools were used but does not identify them.
- 5:26:01
Developers’ Opportunity to Build Products with AI
The speaker argues that development experience combined with AI tools creates strong product-building opportunities, provided the idea starts with market understanding and considers local, regional, or global distribution.
- 5:29:53
How AI Shortened Project Setup
The discussion contrasts the time once spent building boilerplate and initial project structure with the faster setup now enabled by AI tools, in the context of a changing developer job market.
- 5:30:13
Validate Product Economics Before Building
The speaker advises developers to reverse the usual order: study the market, define monetization, and assess margins, pricing, and distribution before building—especially now that implementation can move quickly.
Got a question about something in this stream?
Ask in the group