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
- 0:53:37
Burnish’s Closed Loop and Measuring Action Impact
The discussion frames Burnish as an end-to-end Shopify product connecting research, monitoring, action, and measurement. It also considers linking each execution batch to score-change data for impact attribution, while the speaker explicitly says some measurement concepts and unclear terminology still require deeper understanding.
- 0:57:33
Multi-Persona Agent UX
The speaker describes a Burnish concept that divides the agent experience into tracking, execution, and informing roles, while noting that the idea still needs further development.
- 0:58:37
Minimizing Shopify App Permissions
The speaker discusses requesting a limited set of Shopify permissions as a potential trust advantage because some merchants are uncomfortable granting broad access. He considers it useful, but not a decisive differentiator on its own.
- 1:03:14
Using AI to Understand Study Materials
The speaker argues that tools such as NotebookLM can ingest course material and help a student genuinely understand it, emphasizing learning rather than cheating.
- 1:03:52
Framing Burnish’s Landing-Page Differentiators
The review notes four proposed landing-page angles—truth map, overnight proof, curve, and loop—alongside permissions and measurement, and identifies automatic blog-content publishing as something to avoid.
- 1:05:26
UI Verification and Cautious Progress Reporting
The speaker proposes a UI verification sweep and explains that combining outdated milestones with later-version milestones can inflate a progress indicator. He says V1 is nearly complete and characterizes the remaining work mainly as debugging and iteration.
- 1:06:22
How University Systems Can Undermine Learning Motivation
The speaker argues that perceiving a weak university system and limited impact from a degree can undermine a student’s motivation to learn, and expresses sympathy with that feeling.
- 1:07:29
Tracing an Intermittent 500 Error After Refresh
An HTTP 500 appeared after a hard refresh, while a subsequent reload worked. The discussion calls for checking the logs to identify and fix the intermittent behavior, without claiming that a cause was found or a fix completed.
- 1:10:28
Auditing What Is Actually Wired in the Application
The discussion questions whether shipped-versus-unshipped judgments came from documents or direct code investigation, then proposes identifying the navigation and pages and sweeping the codebase to verify what is actually wired. It frames this as a backend-wiring audit, separate from frontend design work, with findings to precede later project-idea discussions.
- 1:14:11
A Switchable Demo State for the Pulse Page
The discussion proposes fixture data that reveals scores and currently blurred sections as they would appear after a store scan, while first confirming feasibility. The chosen direction is seamless switching between real and demo states, starting with one example and potentially supporting several later.
- 1:18:31
Browser-Based AI Agents for Debugging
The discussion explains how a browser-connected AI agent can inspect a page, network activity, the developer console and logs, capture screenshots, and assist with debugging, though only to a qualified extent. It also describes a workflow where Claude diagnoses an issue and passes a prompt to Codex for the fix, followed by a general comparison of open-source AI browsers and differing levels of Chrome access.
- 1:23:55
Auditing Product Routes and v1 Gaps
A codebase review found many routes absent from navigation, potentially representing onboarding flows, settings pages, or orphaned routes. The notes indicate that some product areas are fully wired while the question-intake interface remains a gap, and they identify customer support as important because the first release is expected to contain bugs.
- 1:28:38
An Activity Log for Burnish’s Background Work
The speaker proposed an in-app area that periodically shows users what Burnish is doing in the background through events or an activity log. He viewed this as a differentiator from competitors focused on content calendars and article generation, and expressed the opinion that FAQs may suit GEO better than blogging, while blogging is more useful for SEO.
- 1:31:33
Keeping Demo Mode Active During App Navigation
The review found that the demo disappears after navigating from Pulse to Mentions and back. The proposed behavior is to keep demo mode active across the application until the user explicitly exits it.
- 1:33:15
Evidence-Based Review of the Mentions Page
For the move to the Mentions page, the workflow emphasized inspecting the actual implementation instead of guessing, using available Shopify development and GraphQL access, and reviewing the page in the browser with screenshots and the established self-scoring system.
- 1:34:19
Refreshing the Website and Optimizing the Landing Page
Some website pages were described as outdated even though parts remained connected, while other areas such as the command center and backend work had changed. Remaining work included polish and an additional landing page, with the call-to-action button identified as a starting point for conversion optimization.
- 1:36:05
Incremental Learning in Web Design
The speaker mentioned using a scoring system to help review the work and observed a substantial difference between current designs and the earliest streams, describing the improvement in web design as gradual learning through repeated practice.
- 1:40:59
Programming Experience and Vibe Coding in the Age of AI Tools
Luqman argues that programmers retain an advantage because technical knowledge can help them go deeper and work faster with AI tools. He cites a published experiment in which, reportedly, the gap between an experienced programmer and a skilled AI-assisted “vibe coder” was not large. In his view, easier software and website creation shifts more attention toward marketing and reaching users; these are opinions and a cited example, not a test performed during the stream.
- 1:44:44
Directing a Coding Agent Without Scope Reduction
After rejecting a time estimate he considered unreliable, Luqman gives a coding agent concrete implementation constraints: complete every requested item without narrowing scope, allow sub-agent assistance, reuse the existing SVG files for AI-engine logos, and place each logo before the engine name.
- 1:47:02
Forking an AI Conversation While Preserving Context
Luqman explains that conversation forking is useful when several directions deserve exploration. Each branch can pursue a separate path while retaining the accumulated context, research, and understanding, avoiding both distraction in the original session and the context loss of starting from scratch.
Got a question about something in this stream?
Ask in the group