Inside the Core 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 an in-development Shopify app intended to improve stores’ visibility on AI platforms, with a capability audit exposing gaps between the published plan matrix and actual implementation—especially APIs, webhooks, and multi-store agency support. Luqman demonstrates a review workflow combining specialist agents for security, compliance, and UX with hands-on testing, human-controlled merges, and automated deployment; mobile inspection revealed visual defects, confirming that the product was not finished. The discussion also covers free-versus-paid report design, A/B testing the placement of email capture, and an audit-led acquisition funnel whose channels should be measured before ad spending. Audience questions explore Claude, Codex, agent skills, and conversation-context compaction, with advice to master reliable tools instead of chasing every new model. On GEO, the speaker distinguishes it from SEO while leaving the effects of AI-generated articles, backlinks, and llms.txt unresolved because the evidence is not conclusive. The practical conclusion is to align public claims with real capabilities, understand integrations, and test the complete experience before Shopify submission; the completion date, best conversion flow, and need for external monitoring remain open.
109 moments from the stream
- 3:03:04
Launching a Shopify App with Genuine Reviews and Early Traction
The discussion rejects fake reviews in favor of Shopify policy compliance and genuine feedback from real users. It also outlines building a lead pipeline before approval so installs can begin early, while acknowledging that outcomes are not guaranteed.
- 3:08:36
Exploring Mobile Access to Coding Agents
An exploratory discussion—whose exact tooling is partly unclear in the captions—about accessing an agent on Android through Telegram, Discord, or Termux, connecting Claude to a phone, and the stated Linux-support limitation for Codex at the time of recording.
- 3:11:42
Specialist Review Before Manual QA
The workflow runs specialist agents over the code and completed milestone, classifies security, UX, and compliance findings, fixes them through pull requests, and only then proceeds to manual walkthrough QA while documenting smaller residual issues.
- 3:15:05
Fixing Critical Issues and Rerunning Auditors
After critical issues surface, the session proposes shipping fixes and rerunning the auditors to verify their application before continuing. The captions later report that the pull request was merged and pushed to main, but do not document a final test in this segment.
- 3:17:23
Reviewing Design Before a New Phase
Before starting a new phase, the speaker asks for a concise rundown, an assessment of the current design, and an explanation of the proposed appearance and implementation approach. The agent is also asked whether design skills such as Impeccable will be used; the feature details themselves are fragmentary in the captions.
- 3:22:50
The Gap Between Backend Work and Design
The speaker felt the backend work was going well while design tasks became confusing, and expected SVG skills to be useful, though he left the outcome open to experimentation.
- 3:27:24
Limiting the Interface Change to the Report
The version looked better because of the faded teaser, but the guidance was to avoid refactoring the entire interface or changing the scan teaser at the top, keeping the work focused on the report.
- 3:34:11
Inline View and Full Report Page
A brief distinction between rendering a report inline and using a link that opens a separate page with the full report.
- 3:37:11
Improving the Teaser Before Email Sign-Up
A review of a faded teaser concept, with a proposal to emphasize AI-engine icons and numbers beneath them so the summary feels clearer and more compelling before asking customers for an email address. The UI step is treated as critical and in need of more refinement.
- 3:39:01
Sequencing Work After a Merge
A discussion of whether to continue immediately with post-merge cleanup and an M11 walkthrough, then rerun specialist reviews to find remaining issues before stepping back to reassess the bigger picture.
- 3:40:20
Visual Quality Review and Core-Engine Gap Tracking
An explanation of recording small defects during a visual-quality walkthrough, completing cleanup, and then cross-referencing the application with the core engine to identify capabilities or metrics that are not represented. Some comparison details are unclear in the captions.
- 3:46:27
Gemini Desktop and Dividing Work Across AI Tools
An initial reaction to a Gemini desktop-app announcement and an apparently substantial CLI update. The speaker was interested in trying the app after release and argued that Gemini may be stronger than Claude for some image- and video-generation tasks, potentially supporting a division of work across Gemini, Claude, and Codex. Product quality and possible pricing remained uncertain because it had not yet been tried and its price was unknown.
- 3:50:31
A/B Testing Two Lead-Capture Flows
A comparison of two marketing-funnel paths: request only the store URL first, then show a result and ask for a business email to reveal the breakdown; or request both the URL and email on the landing page. The speaker noted that delaying the email request could add drop-off, but the better flow cannot be known in advance, so both should be implemented and compared through an A/B test. The broader recommendation was to refine the funnel carefully before spending on ads.
- 3:53:44
Reducing Drop-Off at the Email Prompt
A discussion of how the wording and placement of an email prompt might reduce merchant drop-off. The speaker acknowledged that A/B testing was not currently available and called for a reasoned hypothesis rather than certainty.
- 3:54:49
Coding-Tool Usage Limits and an Impression of Codex
A brief comment about weekly usage limits in a Google plan, with the speaker saying it would be tried, followed by a personal impression that Codex had recently improved. No structured comparison was presented.
- 4:00:42
Prioritizing Fixes and Preparing a Full Application Audit
The speaker proposed starting with a documentation change and a small fix, then moving into a full application audit. Preparation would include checking the capability matrix and pricing page against actual functionality and organizing issues found in earlier reviews, while acknowledging that overlapping milestones had made the scope and references unclear.
- 4:03:33
How Renaming Milestones Can Disrupt Execution
Luqman explains that changing the milestone naming convention and mixing old and new work made the command center difficult to follow. Although it could be reorganized, he prioritized continued execution at the time while acknowledging that the disorganization made milestone references unclear.
- 4:09:16
Auditing the Capability Matrix Before Strategic Decisions
The group discusses comparing the customer-facing plan matrix with what is actually available in the application, then turning identified gaps into execution milestones. The stated priority is to close and clean up those milestones first, followed by another audit to catch anything missed before moving to an application-wide strategic discussion.
- 4:11:14
What Happens When a Long Coding-Agent Conversation Is Compacted?
Luqman explains that long coding-agent sessions may be compacted by summarizing key prior context and carrying it into the next context window. After repeated compaction, early details may fade, so the user may need to remind the agent of the original work and point it back to earlier context.
- 4:14:08
Distinguishing the Gemini macOS App from Coding Tools
The discussion clarifies that the referenced product is the regular Gemini desktop app for macOS, comparable to a general chat assistant interface, rather than a command-line or coding-specific tool.
Got a question about something in this stream?
Ask in the group