Launching New Website + Scan funnel
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 practical Burnish website work: unifying template families and the design system, refining grids, motion, spacing, and the ChatGPT, pricing, Manifesto, and scan-funnel experiences, while planning mobile versions. Several issues remained unresolved, including elements hidden behind the header, scrolling behavior, overlapping voice-over, and unfinished pages; by the end, the site was available but still needed fixes. In parallel, Luqman discussed coding-agent limits, task drift, Command Center and Sentry workflows, and keeping a human in the loop, arguing that workflows should fit the person and task. Audience questions covered learning by building with Claude or Codex, freelancing and Upwork, SaaS distribution, commerce and payments in Egypt and Saudi Arabia, databases and hosting, security, and marketing a medical-booking app. The stream also explored Remotion-based video and synthetic voice-over production, but no final voice choice or proven advertising result was established.
135 moments from the stream
- 2:25:31
Prioritizing Mobile Optimization and the Audit Funnel
Finishing the website, optimizing it for mobile ahead of desktop, and completing the audit funnel were identified as essential launch prerequisites.
- 2:26:41
Discovering New Shopify Apps Through Weekly Roundups
The speaker explained that a third party found newly released Shopify apps by collecting weekly release data and then selecting some for a roundup. Names and figures connected to the private project are omitted.
- 2:27:43
Distinguishing Direct Competitors from Adjacent Tools
While reviewing another tool in a nearby category, the speaker considered its pricing and positioning and expressed the view that it was not a direct competitor despite sharing the broader field. The discussion illustrates comparing product scope, rather than category alone, when assessing competition.
- 2:33:13
A Community for Working and Learning Together
He explains that the community lets members open sessions and work alongside one another on separate projects while sharing knowledge and learning from each other, rather than necessarily collaborating on one project.
- 2:37:08
Building AI Applications for Shopify Merchants
Asked what he does, he describes freelance work using AI to build B2B software and applications on Shopify, and notes the need for a concise, reusable way to explain that work to others.
- 2:40:12
Choosing a Practical Vibe-Coding Path
He advises choosing a path based on goals, experience, and circumstances: vibe coding and AI can support freelance project work, while building a business calls for market research followed by product development and marketing. He qualifies the advice as context-dependent rather than universal.
- 2:42:05
What to Do After Learning Python
Responding to a beginner who has learned Python, he begins by recommending the use of AI tools and moving into practical building, but the answer cuts off before the steps are fully explained.
- 2:42:32
Learning App Building Through AI-Assisted Practice
Advice to move from learning fundamentals such as Python into building a small personal tool with a coding assistant such as Codex, then develop skill by completing practical projects. The discussion emphasizes defining the desired result, learning how to work with AI, and iterating, while acknowledging that this is a simplified account rather than a complete replacement for programming knowledge.
- 2:50:06
Mapping Website Templates to Affected Pages
With another session already working on the homepage, the work shifts to page families and templates, requesting a list of templates and the number of pages each one affects before proceeding with changes.
- 2:51:54
Reviewing the Pricing Page and Calls to Action
A brief website implementation review after part of the work is completed: mobile work is deferred until later, then attention moves to the pricing page and its calls to action, including checking that one is working.
- 2:53:09
Checking Link Routing Before Opening a Pull Request
During implementation, the speaker asks to verify that a link is wired correctly and reaches the intended destination before pushing changes and opening a pull request, then identifies the changelog section.
- 2:55:52
Why AI Coding Workflows Should Fit Their Users
The speaker explains that he no longer relies on one fixed setup and that an effective workflow depends on how each person works and what they are trying to accomplish. He says Command Center helped with building, but he set it aside when it did not keep pace with debugging needs and rapidly evolving AI tools, while leaving open the possibility of improving it later.
- 2:58:41
Agent-Assisted Sentry Triage with a Human in the Loop
The speaker describes a system where Sentry alerts send bugs to a dedicated channel and an agent helps classify each issue and assess its severity. Automated fixes and pull-request merging are presented as a future possibility; for now, he prefers reviewing bugs himself before relaying them to the agent within an orchestrator-and-subagent structure.
- 3:00:09
Command Center Architecture: Subagents and a Connected Task Map
The speaker revisits a workflow built in March and April with an orchestrator and several subagents, which he says helped him build quickly but later became outdated and lacked an adequate debugging system. Its interface uses a swimlane-style project map with subtasks and a task board connected to each node, and he suggests it could be extended and integrated with Sentry.
- 3:03:02
A Human-Supervised Multi-Agent Command Center
An explanation of a command-center design where agents move tasks across board states under human supervision, with an orchestrator and several specialized agents. Agent activity is recorded in a shared JSON file so tools and models can share state; the repository can be downloaded or forked and includes README documentation. A custom MCP connection was described, alongside the concern that maintaining the system at the pace of AI development can distract from the primary project.
- 3:06:49
When a Coding Agent Produces Unrequested Work
The speaker stopped an agent after it spent substantial time producing an unrequested result that he judged unusable, illustrating the cost of agent drift and the need to monitor and interrupt execution when necessary.
- 3:08:09
An Automated Error-Tracking Workflow with Sentry
In response to a debugging-tools question, the speaker described his current workflow: Sentry can capture errors before a client notices them in some cases, then send alerts through Slack or email, with the alert channel connected to a coding assistant to help automate debugging. He said this was his main debugging system and did not identify another tool he currently uses.
- 3:10:05
Desktop Homepage Progress and Remaining Mobile Work
The speaker said the homepage was nearly complete, while the mobile version still remained unfinished.
- 3:16:06
Documenting a Reusable Design Treatment Across Website Sections
While working across several parts of the site in parallel after completing the homepage, the discussion calls for clearly recording the intended design treatment in the document so it can be referenced easily and applied consistently across different sections.
- 3:30:50
Reviewing Website and Ad Launch Readiness
A brief review of the remaining work: the website might be completed that day, while the ads remained a concern.
Got a question about something in this stream?
Ask in the group