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
- 1:49:26
Designing Burnish to Deliver Continuous Store Value
Luqman frames a core V1 question for Burnish: how can the product continuously contribute to improving a merchant’s store, potentially helping address churn? He cites StoreRank’s article-autopilot/content-calendar feature as an example of recurring activity that supports SEO and GEO, then asks how Burnish could expand the subtle daily-activity section in Pulse into a clearer, more continuous contribution.
- 1:53:11
Turning structural gaps into an executable milestone
The structural gaps were to be consolidated into fewer, coherent milestone tasks, each labeled by complexity, then prepared with the project’s preparation command before being handed to an execution agent.
- 1:54:15
Making Burnish’s activity feed the primary home
The home surface was discussed as a clear chronological feed of what Burnish is doing now, helping users understand the workflow and its relationship to the agents. Digest email delivery was deferred, while a persistent in-app ticker remained a future idea needing clarification.
- 1:57:09
Balancing AI-session parallelism with depth
A personal observation suggested that greater AI-session parallelism can speed some backend work but consume more resources, while fewer sessions may produce deeper work. It was presented as experience, not a conclusive benchmark.
- 2:00:30
Managing dependencies and preparing tasks for execution
The milestone workflow was expanded to record dependencies as blockers where needed and update them when tasks finish, while preparing prompts for medium and large tasks before sending them to another agent for execution.
- 2:04:12
Isolating agent tasks with Git worktrees
An automated workflow was described in which each task receives its own Git worktree and branch, followed by pull-request creation and merging after the session. CLI orchestration tools were considered especially useful when coordinating many AI models, though the proposed tool’s identity was unclear.
- 2:06:37
Reviewing confidence and competitor-comparison UI
The page was viewed positively overall, with suggestions to place each model or version in a light box, investigate Shopify Dev MCP capabilities, and left-align the measurement explanation. An awkward spacing issue was also noted in the competitor-visibility section, although the final observation was cut off.
- 2:08:10
Improving Citation Source Breakdown Readability
A gap was identified in the citation source breakdown UI, with a suggestion to give each section its own subtle divider for clearer organization and readability.
- 2:08:31
Session Workflow with Separate Worktrees
The discussion considered creating a separate worktree for each task or session, then closing it once the work is satisfactory and opening a pull request for manual merging.
- 2:09:27
Teasing Features Restricted by Plan
The starter-tier demo exposed results from services that were expected to be unavailable. The proposed treatment was to hide the details while retaining a blurred visual cue, such as an obscured score, to signal the feature without granting access.
- 2:15:25
Planning a Relevance-Driven Pulse Feed
A Pulse concept was discussed that orders relevant events by score and includes tracker, executor, and informer roles. An implementation plan was requested, allowing the component to be repositioned, initially trying all bands and narrowing them later if needed, with Shopify development tooling and skills considered.
- 2:19:16
Coordinating Agents Across Branches and Worktrees
Before execution, a possible clash with another agent working on the frontend design was raised, making the current worktree and branch important coordination details for parallel work.
- 2:20:09
Launch Is the Start, Not the Finish
The discussion frames an app launch as the start of market work: keep pushing and watch for traction and demand before building something else, while preparing for an anticipated seasonal trend in September.
- 2:24:12
Understand Metrics Before Shipping Their UI
The review stresses understanding metrics such as F1, sample size, and measurement logic before shipping, then critiques the construction of a card containing important model and metric information and asks for a clearer visual treatment.
- 2:26:39
Starting a Fresh Worktree from Main
The discussion proposes creating a fresh branch or worktree from main, identifying the conflicts or issues involved, and notes that the home strip needs a “how it works” link.
- 2:28:57
Aligning Supporting Information Below a Score
The correction clarifies that the score itself was not the problem; the subheader and supporting information beneath it looked misplaced and misaligned.
- 2:30:59
Safe Checkpoints During Parallel Development
When parallel changes feel risky, the discussion proposes pausing at a safe checkpoint, committing, and merging after the ongoing migration completes instead of continuing without a clear recovery point.
- 2:32:27
Structuring Metrics Inside a Card
The review proposes consulting the Shopify dev MCP to determine whether a small table or divider can be added inside the card, so the metrics no longer appear to float in empty space and their relationships read more clearly.
- 2:33:28
Securing Branch and Worktree Changes Before Continuing
The discussion focuses on committing and pushing uncommitted branch and worktree changes before reorganizing the work around the next milestone.
- 2:35:28
Finishing the “How It’s Measured” Section
After the card improves, the review asks for one final adjustment: center the “how it’s measured” section for a more consistent presentation.
Got a question about something in this stream?
Ask in the group