Working on Merchant Facing UI
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 preparations for a Shopify app submission, prioritizing completion of the move from Polaris React to Polaris Web Components and visual and functional QA of the merchant interface before returning to website, report, and funnel polish. Practical work also covered GA4, Google Tag Manager, and PostHog configuration; verifying that changes reached production through GitHub and Cloudflare; reviewing Supabase table organization; and troubleshooting MCP access—without treating visible identifiers alone as proof of end-to-end tracking. Luqman drew lessons from accumulated Git branches and worktrees and from coding-agent completion claims that failed manual inspection, arguing for regular integration and direct verification. He also discussed automated build-and-audit agent loops, his personal impressions of Codex and Claude, and audience questions about deployment, model choice, project languages, client acquisition, programmatic SEO, Kanban boards, and Hermes. Open questions remained around some Cloudflare and MCP failures, the best database organization, the value of particular agent tools, and the final submission date.
104 moments from the stream
- 1:38:05
Treating Cloud Task Time Estimates Cautiously
While monitoring a cloud task, he observed that its actual completion time could differ substantially from the displayed estimate, so he treated the timeline as approximate.
- 1:40:04
Reviewing Measurement Configuration Before Publishing
He considered whether to defer Google Tag Manager and continue optimizing the analytics setup, while stressing that visible measurement identifiers should be removed before making the material public.
- 1:42:25
Configuring enhanced measurement and analytics environment variables
The setup progressed through GA4 measurement, including enhanced measurement and events, before moving to three previously created environment variables. The discussion also noted that the GTM section required additional account setup that might be better completed off-stream while work continued on the main project.
- 1:47:22
Choosing manual Cloudflare sign-in during setup
During Cloudflare setup, after an option appeared that would let the browser interact with any website without asking, the speaker chose to open the Cloudflare site and sign in manually. No account details are included in the index.
- 1:51:29
Finding Web Design Clients and Using Upwork Carefully
A discussion with an audience member about marketing ownership and client acquisition for web design services, including a suggestion to explore demand in Gulf markets. Luqman also cautioned that violating Upwork’s policies could, in his understanding, lead to an account ban.
- 2:01:05
Breaking a Large Milestone into Tasks for a Coding Agent
Luqman described preparing an important Shopify migration milestone by dividing it into tasks, writing prompts, and monitoring the agent’s progress against its estimate. Private work and usage figures are omitted.
- 2:01:53
A Lesson from Accumulated Git Branches and Worktrees
Luqman explained that working across many branches and worktrees for a week without regular merges or commits created a confusing backlog. Merging the changes then broke the code and exposed an old placeholder site; after recovering, he framed it as a practical lesson in branch and worktree management.
- 2:03:39
Triggering a Cloudflare Deployment after Setting Environment Variables
Luqman tried to help another coding agent trigger a Cloudflare deployment after environment variables had been added. He clarified that the goal was to trigger the deployment itself, not necessarily deploy the landing site, while investigating the correct route for doing so.
- 2:08:47
Codex Integration with Supabase through the Plugin
Luqman said they were re-enabling the Supabase MCP integration despite having configured it before, and observed that installing the Supabase plugin in Codex helped set it up. This was a practical observation, not a completed comparative test.
- 2:09:22
Qualified Impressions of Cloudflare Workers, Vercel, and Railway
Luqman said he uses Cloudflare Workers while acknowledging limited experience with both it and Vercel. He expressed a general technical preference for Vercel alongside a personal objection to using it, and attributed to another person an uncertain view that Railway may suit certain application types better, stressing that the platforms are not identical substitutes.
- 2:13:27
Codex Plugins and Project Service Configuration
Luqman explains that Codex was not reading some global project settings as expected and suggests plugins may be the intended way to connect MCP and service configuration. He says the Supabase plugin worked immediately, then tries a Cloudflare plugin.
- 2:16:08
From Measuring Merchant Visibility to Improving It
Luqman discusses expanding the tool beyond measuring merchant visibility on AI platforms so it can recommend best-practice improvements intended to raise the score, based on his GEO research. He connects this work to the planned merchant-facing interface.
- 2:18:06
Zero Data Retention and Model-Provider Selection
Luqman explains that Shopify review may require clear privacy and terms-of-service details for model providers. He says he checked zero-data-retention availability, but DeepSeek required additional contact and might attract extra scrutiny, so he chose to defer it; he also gives his opinion that other platforms currently matter more for merchant product visibility.
- 2:19:48
Repeated Cloudflare Deployment Failures
Work is held up by Cloudflare deployment attempts that take a long time and repeatedly fail, leading Luqman to ask Codex to perform the deployment directly. The cause of the failure is not established on this page.
- 2:22:46
Deploying from GitHub and comparing Codex with Claude for coding tasks
Responding to a question about connecting GitHub to deployment services, the speaker explained that Railway was linked to the repository while Cloudflare might require an additional setup step. Based on one deployment task, he said Claude completed it faster and seemed to understand his coding context better, while stressing that Codex may be stronger at other general tasks and that these were personal observations rather than a universal verdict.
- 2:25:35
Choosing a lower-token-use coding model when you have technical experience
The speaker said he had not tried fast mode. Separately, he said he could not get a multi-agent experiment running and cautiously recalled possible hallucination concerns. He added that someone with programming experience may be able to rely on an older or less capable model for longer and use fewer tokens.
- 2:27:21
Reading coding benchmarks and choosing models by task
The speaker reviewed a coding section on an official launch page and compared its results with his hands-on impressions. He felt the displayed coding benchmark showed only a limited improvement between two successive releases and a larger gap versus Claude, while acknowledging that he did not know the benchmark’s exact methodology. He also argued that model choice depends on the task: he preferred Claude for building and coding, while regarding Gemini or higher-reasoning modes as more suitable for academic research, explicitly as his own assessment.
- 2:30:41
Project languages and limited coding experience with Gemini
Answering an audience question, the captions identify JavaScript, Python, and TypeScript as languages in use. The speaker noted that he does not come from a coding background and was unsure about some details of how the agents implement the code. He added that he had not used Gemini extensively for coding, though his limited experience with it in Antigravity was positive.
- 2:32:03
Auditing Polaris Web Components Usage
A focused audit was proposed to answer a specific question: whether the application UI fully uses Polaris Web Components rather than Polaris React.
- 2:33:02
UI Work as Part of Building a Shopify Application
The discussion clarifies that user-interface work is part of developing a Shopify software application, rather than a task detached from the broader application engineering effort.
Got a question about something in this stream?
Ask in the group