Debugging Core Product Engine
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 moving Burnish toward a Shopify-ready V1: reviewing its product engine and value proposition, auditing what is actually wired between the frontend and backend, and refining onboarding plus the Pulse, Mentions, and settings sections around Polaris and Built for Shopify expectations. Luqman discusses making background-agent work visible through a continuous activity feed and connecting research, action, and measurement in a closed loop, alongside demo-state and free-audit funnel ideas; whether these ideas are sufficient, and how attribution and measurement should work, remain unresolved. Practical work includes tracing intermittent HTTP 500 errors and missing CSS registration on the Pulse route, noting an initial Supabase connection problem and caution around resetting onboarding state in the database without establishing a diagnosis, and auditing branches, Git worktrees, and pull requests after mismatched development environments made preserved changes appear lost. Audience questions and side discussions cover context-dependent Shopify app selection, SEO/GEO and product metadata, browser agents and Codex/Claude workflows, broader Git and agent-orchestration guidance, evaluating agent skills, AI-assisted study, and the tradeoff between livestream latency and quality. Other exploratory discussions address Product Hunt’s suitability for Shopify apps, traffic measurement, advertising, and possible discrimination around product launches, as well as practical comparisons of AI-generated storefronts; some model names and versions are inconsistent in the captions, and perceived speed was not formally measured. The stated direction is to finish Burnish’s core, simplify onboarding, and conduct a thorough Shopify compliance review before expanding marketing, while acknowledging that limited development-store testing, expected V1 bugs, migration and performance issues, and several measurement questions still require verification and concrete resolution.
121 moments from the stream
- 4:43:24
Diagnosing Missing Styling on the Pulse Route
A comparison with a screenshot made the Pulse section appear broken despite assurances that the code remained. The speaker requested concise directions for another agent rather than a direct fix, and the issue was later attributed to the interface CSS not being registered on the Pulse route.
- 4:48:42
Reviewing Built for Shopify and Polaris Compliance
The restored interface deliberately followed the existing Pulse component pattern, but it did not meet the expected Built for Shopify and Polaris Web Components requirements. The response said Shopify development tooling had been used but that its verification was misleading, so a prompt was requested for another agent to rebuild the interface compliantly.
- 4:51:16
Choosing Shopify Apps Based on Store Context
An audience member asked which Shopify apps could help improve conversions and sales. Rather than giving a generic list, the speaker said recommendations first require context such as the business’s size and location.
- 4:55:21
Scanning a Shopify Store for SEO and GEO Visibility
The speaker demonstrated an ongoing trial scan of a Shopify clothing store, examining AI-answer visibility and selected product fields. Images and descriptions appeared generally acceptable in this pass, while fields such as alt text and tags were absent; he said these factors matter to both SEO and GEO. He also found that a broad comparison query needed geographic context such as Egypt or the Arab region, while stressing that the analysis was still in progress.
- 4:56:48
A Dated Observation on Shopify Sidekick’s Evolution
The speaker said he personally used Shopify Sidekick during September, October, and November 2025 and knew its capabilities at that time. He expressed surprise at how much it had advanced between November and March, attributing that change to OpenAI as his own opinion.
- 5:00:27
Preparing for Shopify Release and Checking Milestone Task Updates
The discussion mentions needing beta users before publishing on Shopify, then identifies that tasks do not appear to update inside a milestone and points to work associated with an onboarding rebuild.
- 5:02:51
Clarifying the Scope of a Shopify Operations Manager Role
The discussion explains that “Shopify Operations Manager” is a broad title, so useful guidance first requires clarifying whether the work concerns supply chain, online-store operations, or another concrete responsibility.
- 5:07:59
Pausing After the Milestone Because of a Migration Issue
With two tasks remaining in the milestone, the plan is to stop once it is complete because the implemented work still has a migration problem; the captions do not provide enough technical detail to identify its cause.
- 5:08:45
Suggestion to Try a Frontend Tool and Compare It Side by Side
A participant recommends trying a tool or version whose identity is unclear in the captions, describing it as currently free, fast, and powerful especially for frontend design. A side-by-side comparison is then proposed, but the comparison targets are not clearly established.
- 5:10:59
Comparing Two Models on a Sunglasses Landing Page
Luqman ran a side-by-side LLM Arena comparison between models he identified as Qwen 3.6+ and GLM 5.1, giving both the same prompt to build a sunglasses e-commerce landing page with React and Tailwind CSS. He discussed making the prompt specific while leaving room for design choices, then struggled to preview the React output before locating the build-apps-and-websites option. In the displayed results, Qwen finished first, though he criticized weak content and a missing clear button in one preview, while GLM's output appeared poorly connected to sunglasses. His preliminary judgment favored Qwen because it at least included a relevant image; the comparison remained exploratory.
- 5:28:38
Balancing Stream Latency and Quality
Luqman explained that reducing stream latency could make live conversation easier, but may lower stream quality.
- 5:31:15
Comparing AI-Generated Storefront Designs
A practical review of storefront outputs from multiple tools. The designs resembled Shopify stores, and one reviewed design included basic information, testimonials, and conversion elements, while some links were broken and one output appeared to omit an unclear element. The speaker considered the GLM and Quinn results broadly similar and tentatively preferred Quinn for its perceived speed, while explicitly noting that completion time was not formally measured and that images could be handled later.
- 5:34:56
Observations on a Model-Comparison Tool
A brief discussion of a comparison tool that seemed useful for occasional checks, noting that some models were unavailable and expressing a personal preference for another option because of its speed.
- 5:36:43
Why Did Starting Shopify App Development Become Complicated?
While trying to start a Shopify app development environment, Luqman questioned why a normally simple one-line task had become difficult, attributing the problem to overanalysis and overcomplication.
- 5:38:22
Checking Merge Safety, Conflicts, and Automatic Post-Merge Verification
Before merging a pull request, Luqman checked whether the session’s work could be lost after conflicts appeared across several files. He observed that the tool detected the conflict and unexpectedly created its own loop to monitor the merge and verify the post-merge state, despite not being asked to do so. After merging and squashing, he confirmed the work remained and moved on to further refinement.
- 5:47:02
Resetting the App for Visual Onboarding QA
The application’s onboarding state was reset in the database to return to the first welcome step, enabling visual review of the sequence from the beginning, with caution noted about possible Supabase issues.
- 5:53:51
Why Onboarding Must Be Fast and Simple
The discussion argues that a lengthy onboarding flow may cause unfamiliar incoming users to leave or uninstall. It proposes a seamless, fast, simple experience with scanning performed in the background, then requests diagnosis of a delay lasting several minutes while treating the development server only as a possible cause, not a confirmed explanation.
- 5:57:15
Matching Coding-Agent Effort to the Task
The speaker explains that reasoning effort should vary by prompt: raise it for important technical work such as backend debugging, and lower it for general questions or directional guidance, based on the task’s importance and demands.
- 5:59:33
Orchestrating Multiple Tasks with Modes and Custom Commands
The speaker describes a personal workflow that prepares prompts for a batch of tasks and then runs a custom automation command through a command center, switching between automatic and planning modes according to the task.
- 6:05:11
Simplifying the GitHub Workflow and Using It as a Source of Truth
The speaker reflects on how branch and worktree complexity disrupted momentum and says the workflow needs simplification. He considers learning GitHub before launch valuable for reverting changes and reducing the risk of treating a local folder as the sole source of truth, while acknowledging its learning curve.
Got a question about something in this stream?
Ask in the group