Days From Shopify
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 developing a Shopify app for improving merchants’ visibility in AI answers, including reviews of the reputation and discovery sections, presence-score design, intent-based query ranking, and clearer presentation of answers and research. Luqman argues that measurement should use web-grounded research across multiple engines, while provider quality, cost, and policy-compliant access must be verified; the scoring formula, engine selection, parts of the product architecture, and pricing remain unresolved. The work also covers Market Scout planning, an activity log and dashboard, onboarding, quality checks, and Shopify submission preparation, alongside inconclusive troubleshooting of Shopify and Claude outages and YouTube chat auto-translation. Audience discussions include GEO versus SEO, AI coding-tool limits and pricing, livestreams as shared workspaces, professional networking, and cybersecurity careers, while distinguishing responsible vulnerability disclosure from harmful or unlawful suggestions.
59 moments from the stream
- 1:22:27
Preparing for Shopify App Submission
The speaker describes the expected shift after submitting an app to Shopify: from building toward maintenance, marketing, and customer acquisition. Some questions—such as ad performance and cost per install—remain unknowable until the market responds, while error tracking, automation, and email flows are part of launch preparation.
- 1:27:32
Observations on Token Usage and Model Cost
Based on personal experience and comments from others, the speaker feels token usage has improved somewhat, while relaying an uncertain observation that some Chinese models may consume more or have become more expensive. The exact names and comparison are unclear in the captions.
- 1:31:53
Designing Search Suggestions in the Discovery Section
A discussion of where the pills beneath a search query come from, including catalog categories and topic labels. The speaker proposes using a general label such as “category” instead of listing subcategories, then documenting the discovery section and its three lanes as a single source of truth before continuing the UI design.
- 1:39:14
An Interactive Activity Log and Dashboard for App Transparency
The speaker describes an interactive log that would show merchants what happened inside the app and what actions the app performed, noting that the volume of events requires some structural planning. A dashboard would follow later to connect the application’s parts into a coherent overview.
- 1:41:09
Pre-Submission Quality Checks for a Shopify App
The plan includes a quality-assurance sweep, a mobile-view review, and a Shopify compliance check. The speaker expects Polaris components to simplify responsive work, followed by a Shopify check before formal submission and a short explanatory walkthrough video.
- 1:42:05
Preparing Onboarding and the Website Funnel Before Launch
A review of remaining pre-launch work, including improving the application's onboarding, refreshing the website's design and content, and refining its audit funnel.
- 1:43:08
Build the Core Engine Before the Surrounding Experience
An engineering lesson from starting with a skeleton and moving too early to the website: establish the product architecture and core engine first, since a later application rebuild can leave earlier work misaligned with the new direction.
- 1:44:34
Running an Agent Task in the Background While Working in Parallel
A workflow in which an agent session handles a relatively long-running task in the background while work continues on another priority section instead of waiting for completion.
- 1:48:45
Designing the Market Scout Experience Before Building the UI
A request to expand the design from a single widget into a complete Market Scout concept covering workflow, cadence, overview cards, and how scheduled activity should be represented. Its proposed value is researching beyond the application's internal data and bringing ideas back into it, with brainstorming and concept testing needed before committing to the visual design.
- 1:51:14
Tracing an Unexpected Evaluation Result in the Database
The beginning of an investigation into a job that was expected to be wired into the system but produced an unexpected database evaluation result. The discussion raises whether the data is mocked and calls for identifying the cause, but the page ends before a conclusion is reached.
- 1:55:06
Tying Pricing to Margin and Research Costs
The proposed price remained undecided because it depended on gross margin and the cost of running research, with the possibility of using a different provider or model.
- 1:57:17
Running a Batch Evaluation Across Multiple Engines
A proposal to clear earlier answers, send the question set to several engines in one batch, and then inspect how the system and result flow are wired after responses arrive.
- 2:00:28
An Always-On Agent That Suggests Research Directions
A concept for an in-app agent that monitors data changes and progress, compares them with prior activity, and shows its reasoning to the merchant before suggesting topics or areas for further research and focus.
- 2:03:43
Reviewing Scan Results and Preparing a Merchant Sample
He asks for a visual preview before locking an option, wants to inspect the results of a multi-engine, multi-question scan once it finishes, and considers trying a merchant sample.
- 2:11:17
A Shopify App for AI Visibility Optimization
He explains that he is building a Shopify application focused on AI visibility optimization, describing AEO as a newer domain similar to SEO.
- 2:15:56
Measuring Whether Engines Detect a Business Closure
The speaker proposes scanning a set of questions and producing a table showing which engines reported that the business had closed, how often they did so, and which engines omitted that information. The goal is to compare whether the engines detected a business-closure fact the speaker considered important.
- 2:20:21
Livestreams as Shared Workspaces and the Challenge of a Busy Chat
The discussion describes how background music and conversation during a stream can create the feeling of working alongside someone else. It describes the atmosphere as relaxed with a small audience but anticipates that handling the chat would become harder if participation grows.
- 2:23:53
Comparing Web Grounding Across AI Models
In a displayed research example, the speaker considered Perplexity’s answer more connected to current information, while other answers appeared to rely on model memory. He inferred that the research engines in this context should use the web, while the reason for the difference between models remained uncertain.
- 2:26:56
Building a Professional Network and Portfolio on LinkedIn
The speaker advised starting early on LinkedIn because a professional network takes time to grow, and described the platform as a useful place to present a portfolio even when it brings no immediate return.
- 2:29:11
Requiring Web Search for Reputation Research
The speaker decided that the reputation section needed search enabled for every model, accepting potential added cost and latency provided the search produced useful results.
Got a question about something in this stream?
Ask in the group