TOFU Acquisition Strategy
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 documents wide-ranging practical work on Burnish, a GEO application for Shopify merchants intended to measure store and product visibility in AI answers and recommend catalog improvements. The session focused on rebuilding and simplifying onboarding, polishing the scan page, report-wait experience, and mobile UX, and reviewing plan selection, Shopify billing, and data-processing consent. In parallel, it developed a Meta/Facebook acquisition funnel built around a free report and signals such as scan started, scan completed, and lead, with a recurring view that the mid-funnel and user journey should be improved before ads run. The work also covered clean-state testing and troubleshooting 500 errors, scan failures, deployments, Turnstile, tracking events, Stripe redirection, and Supabase, Cloudflare, Railway, and GitHub integrations; several causes and outcomes remained unresolved, and end-to-end verification was not shown as complete. Performance testing found that reducing sampling depth did not deliver the expected speedup in one attempt, leading to discussion of progressive results and a clearer waiting experience. Audience questions and broader discussions covered home-hosted databases, building Shopify apps through vibe coding, learning programming with AI, GEO versus SEO and AEO, dropshippers as a possible audience, and personal comparisons of Codex, Claude, and agent workflows. The indexed topics also cover separate discussions of Shopify-app market economics and saturation, D2C and Egyptian market strategy, cash flow and AI-tool spending, internet and OBS troubleshooting, digital-access constraints, digital-nomad visas, and asynchronous communication. The supported conclusion is that clear value, low-friction onboarding, reliable funnel behavior, and sound measurement should precede advertising scale; final product readiness, several bug causes, and the best landing-page and consent structure remained open.
183 moments from the stream
- 10:51:25
Building a Shopify App Front End with Vibe Coding
Answering an audience question, the speaker says he is using AI and vibe coding to build a website front end associated with a Shopify app despite having no prior programming or web-development background.
- 10:53:05
Learning by Building
The speaker argues that building a real product is an effective way to learn web development, while noting that practical progress does not imply complete expertise.
- 10:55:41
Balancing Visibility-Scan Speed and Coverage
The speaker explains that reducing the question variants from three to one lowered processing time and cost, and says he believes one variant provides sufficient coverage.
- 11:01:31
Designing the Report-Wait Experience
After observing a faster scan, the speaker suggests encouraging the merchant to remain on the page with a clear time estimate instead of building an unnecessarily complex progressive-report experience.
- 11:03:15
Screenshot-Based Review of an Ad Landing Page
The speaker critiques overcorrecting for a mobile-first requirement while neglecting desktop design and visual details. He proposes opening the page, capturing every page, and reviewing the screenshots before planning improvements because the page is the gateway to the advertising funnel.
- 11:07:14
Improving Report Progress and State Consistency
The review calls for tying the progress bar to real processing stages, merging overlapping stages, removing unexplained labels, and keeping indicators consistent between the summary and full report. It also flags platform states that remain queued or display inconsistent results.
- 11:14:10
Design Direction for Burnish’s Paid Landing Page
The proposed direction for the paid landing page was to study Burnish’s existing homepage, product-explanation page, and pricing page, then carry their visual style and design philosophy into the new page.
- 11:15:24
Advice for a Young Learner: Use AI to Learn
In response to a question from a 17-year-old, the speaker recommended using AI as a learning tool across subjects to accelerate learning. He qualified this by saying it was not advice to abandon school or university, but to recognize that education is changing because of AI.
- 11:17:04
Separating Ambiguity from Real Defects in Agent Debugging
While reviewing a stalled task, the speaker said two of three flags came from ambiguity and one was a real defect, then returned the task to in progress with feedback and three items to patch. He later said Codex could not get past a bug while Claude was working on the task; no final resolution was documented.
- 11:18:06
Automation and Agentic Systems as Promising Skills in 2026
The speaker argued that automation and agentic systems were among the most promising areas in 2026, citing market direction and investment by major accelerators and investors in companies that build for or apply AI. This was presented as an opinion and career recommendation, not a certain forecast.
- 11:20:15
Is Programming Still Worth Learning in the AI Era?
Luqman advised continuing to learn programming when the goal is to work with or through AI tools, while expressing doubt that a long traditional study path fits the direction the field is taking.
- 11:23:27
A Community Connecting Coding, Shipping, and Marketing
Luqman described an Egyptian community spanning development, web design, and business, alongside an ambition to build a broader Arab-world vibe-coding community. He emphasized combining product building, shipping, and marketing, and treating slow movement as progress.
- 11:26:35
Quick Audit and Interface Layout Feedback
He said the shared Markdown audit could not be opened, then requested swapping the interface sides so the form panel sits on the right and the preview panel on the left; the remaining sections were approved to proceed.
- 11:29:44
A Practical Codex–Claude Workflow Comparison
In a personal assessment, Luqman said Claude better infers intent and moves work forward, whereas Codex needs explicit prompting to be proactive and may overdo it when that behavior is configured globally. He qualified that experiences vary, considered Codex good for backend work, and suggested combining both tools when possible.
- 11:33:22
Matching Reasoning Effort to Task Difficulty
Continuing the comparison, Luqman argued that some coding assistants require close supervision and that stronger reasoning modes are better reserved for unusually difficult or emergency-level problems rather than routine use.
- 11:37:13
Improving Onboarding After Debugging and Competitor Analysis
Luqman recapped work on a Shopify app’s onboarding system, explaining its relevance to churn. After finding bugs and debugging them, the team studied direct competitors’ onboarding in depth and reorganized the planned refactor.
- 11:38:24
Polishing Onboarding and the Conversion Funnel Before Ads
An update on parallel work toward a new onboarding system, wiring Facebook Pixel infrastructure, defining custom conversion stages, and polishing the funnel’s mobile and PDF experiences before considering it provisionally ready for ads.
- 11:39:40
A Personal View of Gemini CLI and Antigravity for Coding
The speaker gives a negative personal assessment of Gemini CLI and says that, for coding, he would use Gemini through Antigravity.
- 11:40:54
Seeing Progress by Comparing Work Over Time
The speaker explains that revisiting earlier recordings or states of the work helps reveal progress and restore motivation, even when long periods focused on one area—such as the website rather than the application core—temporarily make advancement feel slow.
- 11:42:01
Monitoring Long-Running Automated Coding Tasks
A brief discussion of automated coding tasks that can take a long time, especially when given a large prompt, with progress monitored before assessing the remaining work.
Got a question about something in this stream?
Ask in the group