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
- 7:55:01
Delegating an Application Refactor and Onboarding Milestone to a Sub-Agent
A practical description of giving a prompt to a sub-agent for a milestone containing several medium-to-large tasks and architectural work, including an array refactor, the onboarding flow, and a scan process. The excerpt describes the intended scope, not verified completion.
- 7:56:56
How Should a Stream Address Its Audience in Arabic?
A brief discussion of an Arabic alternative to “chat” when addressing a live-stream audience. The speaker finds “look, everyone” comparatively long and observes that some Arabic speakers use the English word “chat” directly.
- 7:57:12
What Is a Digital Nomad Visa?
An audience question is followed by a general definition of a digital nomad visa as a permit allowing a remote worker to live legally in a foreign country while working for companies or clients elsewhere, with several countries mentioned as examples. One garbled word in the captioned definition is not relied upon.
- 8:02:22
The Hidden Cost of Digital Access Barriers
The speaker argues that location-dependent access limits, power and internet problems, and IP restrictions force workarounds such as VPNs, consuming time, effort, and mental bandwidth that could otherwise go toward productive work.
- 8:06:20
Correcting the TalkStore Password-Protection Diagnosis
After challenging the tool’s diagnosis, the speaker checked and acknowledged that TalkStore was still password protected. He noted that this roadblock had previously been bypassed, but why it had returned remained unclear; no final resolution is shown in this passage.
- 8:08:37
Uneven Fiber Availability and Upload-Speed Constraints
The discussion broadly contrasts an audience member’s unlimited fiber connection in Algeria with fiber being available only in some parts of Egypt, then identifies upload speed as the weak point in the connection being examined.
- 8:10:13
Diagnosing a Stream Freeze During an Internet Speed Test
The speaker explained that the stream relied on a fast wired connection, then observed that running a speed test coincided with the stream freezing before the connection recovered.
- 8:16:58
Comparing Codex and Claude Desktop with Warp
The speaker said he liked working with Codex and Claude Desktop, particularly their display of branches and trees, and plans to retry Warp and compare the tools again.
- 8:18:06
Starlink Availability in Morocco and Its Estimated Cost
The speaker noted Starlink's entry into Morocco and considered it expensive, citing an explicitly uncertain monthly estimate.
- 8:26:04
Preventing Repeated Agent Stops During Smoke Testing
The speaker wants the agent to complete the entire milestone before returning, after observing that design-related work stops even when the auditor reports a pass. During smoke testing, he proposes recording the reset and page-update fix and understanding why the issue recurs so each test does not incur a long delay.
- 8:31:18
Distinguishing an Opus Frame Issue from a Network Drop
After initially suspecting the network, the speaker concludes that the interruption appears related to Opus frames, while the source of the problem remains unknown.
- 8:33:49
Possible Ambiguity in a Storefront Password Error
The speaker says the backend was reporting a storefront-password rejection, but suggests the message may represent more than one internal failure and could appear even when the password is accepted. The explanation ends before the alternative is completed.
- 8:42:35
How the Store’s LLM Visibility Report Works
The workflow identifies the store’s niche, reads its catalog, derives likely queries, and compares results across multiple language models. The report explains its results and scores and suggests catalog improvements, while score consistency, design quality, and mobile optimization still need attention.
- 8:45:07
Local Model Hardware and Hybrid Market Research
The group discusses choosing a local model suitable for Mac hardware, though the exact model name and specifications are unclear in the captions. The speaker also says market research combined manual and automated work, while audience targeting remains at an early stage.
- 8:46:03
Reset Reliability During an Onboarding Refactor
After a successful run, the speaker asks for confidence that resetting the application will not reintroduce storefront-password and Cloudflare Turnstile problems. This is happening alongside a full onboarding-process refactor, making the change substantial and slow.
- 8:48:55
Locating the Report-Generation Bottleneck
The speaker asks why report generation takes several minutes and how the delay can be measured internally. The initial observation is that most of the wait occurs while building the deep report or preparing its PDF, rather than during the core analysis itself.
- 8:53:20
Configuring Codex with Instructions and Plugins
A practical Codex discussion suggests that detailed instructions and specialized plugins, including Supabase, Cloudflare and Railway integrations, may help guide the agent. However, slowness remained a problem in the experience described even with high-performance settings.
- 8:56:08
Reducing Product-Catalog Scan Time
The discussion examines a scan that caps how many catalog products it processes and notes that runtime does not necessarily scale linearly with product count. Proposed improvements include reducing sampling depth and presenting an initial readable report while deeper evidence is completed later; these are implementation ideas under consideration, not demonstrated results.
- 8:56:44
Allowing Time to Learn Agent Tools
The group discusses how some agent tools, including Hermes, have a learning curve, while a short period of hands-on use may be enough to understand their workflow. These are personal impressions rather than a measured comparison of tools.
- 9:00:09
Potential Roles for AI Agents
The speaker outlines a future vision for an AI agent that could assist with research, customer service, social-media marketing and software debugging. Some of these uses are explicitly framed as future possibilities, not current needs or proven capabilities.
Got a question about something in this stream?
Ask in the group