PROJECT & VISION UPDATE
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 Talk Store’s change of direction after the speaker says Shopify did not approve the permission needed for its original conversational storefront-editing concept. He therefore reprioritized a faster submission-ready version and deferred broad marketing. For future products, he proposes testing several ideas through landing pages and real demand signals—such as sign-ups, reservations, and conversion metrics—before investing heavily in development; this is a proposed method, not a completed validation result. The practical work demonstrates a multi-agent workflow that turns research into tasks and coordinates building, review, security, and validation, while tracking progress and distinguishing genuinely parallel work from blocked dependencies. Although the speaker believes this can accelerate development, he stresses rapid usage-limit consumption, the need to challenge model estimates, and the importance of human review and UX polishing after agents report completion. The stream also examines a prototype agent that analyzes Shopify product data, proposes enrichment, and requests user approval before making changes. The discussion covers rule-based and pattern-matching quality scores rather than AI scoring, with attribute weights varying by product type; it also recommends choosing suitable model quality before optimizing local-inference costs at scale. The confirmation gate is shown working, but deployment problems, failing actions, and a pipeline that might fail silently require log inspection and code-level investigation rather than guesswork; security review, practical testing, and final polish were not consistently complete. Shorter discussions cover ROI-based hardware choices, coding-tool limits, Soundcore headphones, live-viewer behavior, and the pull of short-form video. In a side discussion, the speaker described Egypt’s marriage “qa’ima” custom as unusual and expressed the opinion that rising divorce has other causes not necessarily tied to it, while deferring a fuller discussion to another stream.
71 moments from the stream
- 0:01:29
Talk Store Changes Direction After a Shopify Constraint
The speaker explains that Talk Store’s proposed core advantage was an in-Shopify agent that could modify a storefront through conversation, but, in his account, Shopify did not approve the required scope. He therefore decided to finish a faster submission-ready version, defer the broader marketing effort, and stop treating it as his main project.
- 0:04:35
Validating Software Ideas With Demand Signals Before Building
The speaker proposes selecting several strong software ideas, creating landing pages with clear explanations, and marketing them with sign-up or reservation options before building the products. Demand signals would then be compared over a short period, with development focused on the idea showing the strongest traction. The aim is to reduce wasted effort and exposure to external blockers while letting market demand guide what gets built; this is a proposed method, not a completed validation result.
- 0:12:04
Testing an Idea Before Building and Updating the Roadmap
The speaker explains that a storefront blocker changed the plan and that he intends to follow a playbook for testing an idea and its market before investing heavily in the build. He consequently reprioritized backend and application UI work while postponing some pre-launch marketing work.
- 0:14:33
Reviewing Model Estimates Instead of Accepting Them at Face Value
The speaker argues that language-model estimates require scrutiny and informed pushback. He says technical research and observed progress led him to request a more accurate timeline and revise the plan.
- 0:17:23
A Multi-Agent Roadmap Sweep Before Execution
The speaker assigns reviewer, explorer, researcher, and validator roles to inspect the codebase, roadmap, and swim lane. Their goal is to compare the current state with planned tasks and identify missing work before execution begins.
- 0:19:17
Identifying Tasks Suitable for Parallel Building
The speaker explains that faster delivery requires checking task overlap first: conflicting tasks should not run together, while independent tasks can be executed in parallel.
- 0:21:05
Building Software Under External Platform Constraints
A reflection on the broad possibilities of software development versus rules imposed by app stores, hosting, and distribution platforms, with an interest in models that depend less on external gatekeepers. The discussion is exploratory, not a final decision.
- 0:22:48
Validating Demand Before Building an App
The speaker proposes validating the market before coding by launching the app’s offer and testing marketing response, rather than investing development time without strong evidence of demand. He also gives the opinion that broad Facebook targeting may suit such experiments, while Kickstarter is generally less compelling for software.
- 0:29:24
Updating Tasks and Repairing Project Dependencies
A practical work review identifies several actions: update materials for the correct day, fix broken dependencies, add missing tasks, revise the submission checklist, and review weekly tasks to create a parallel mode. The project identity and some surrounding context are unclear in the captions.
- 0:33:25
Testing Demand Before Building the App
He argues that pre-build market validation can be a valuable investment: test several ideas the founder genuinely believes in, then compare signals such as click-through rate, conversion rate, and cost per install or sign-up (CPI), as described in the captions. He suggests that a small paid reservation may indicate stronger intent than email collection alone, while the results can help assess the idea’s potential before building the app.
- 0:35:33
Why AI Brainstorming Alone Is Not Enough
He shares the view that crowded markets and the uneven quality of ideas produced by carefully configured AI tools make brainstorming alone insufficient. His proposed response is to select a small set of ideas he genuinely supports and test them against the market, while acknowledging that all of them could still fail.
- 0:38:26
Marking Parallel Tasks Precisely
While reviewing a weekly plan, he notices that the parallel-work marker appears on more tasks than the accompanying description supports. He asks for the marker to identify precisely, week by week, which tasks can truly run together without implying that every task is independent.
- 0:39:45
Reframing Project Setbacks
He directs late arrivals to the stream introduction for context on important changes and setbacks, and shares his view that something initially appearing negative may later prove beneficial. The thought remains incomplete at the page boundary.
- 0:40:21
Documenting Project Pivots and Dated Decisions
He explains the value of recording project updates and pivots in a dated reference so later changes and their rationale remain traceable, and mentions possibly making a separate update video.
- 0:41:42
Organizing Blocked Tasks and Dependencies
He reviews the work center so genuinely blocked tasks appear in a clear section with blocking reasons and released dependencies. The goal is to execute tasks in the correct order and apply the organization consistently each week.
- 0:47:21
Inspecting a Shopify App and Defining Its Next Capabilities
Before starting the task, he frames a question to inspect the app’s current state when running the Shopify development environment. He says an earlier direction involving theme duplication and editing was not accepted, so the focus shifted to a capability for working with a merchant’s product section; he cautiously adds that he believes he confirmed this direction is acceptable to Shopify.
- 0:51:27
A Command Center for Multi-Agent Development Workflows
The speaker explains a configurable workflow that turns agent research into detailed task files, passes them to a builder, and records steps and timing. Separate review, security, validation, and compliance roles then inspect the result, with status and per-agent details available for monitoring. He says the approach greatly accelerated the demonstrated work, while noting its high token usage and the need to review agent questions and provide operator judgment when required.
- 0:55:50
Scoping a Universal Audit Trail
The demonstration considers whether a database change should be included in the current migration or deferred to a later billing task. The response explains that the table is intended as a universal audit trail covering both product and storefront actions, and the speaker chooses to follow that recommendation.
- 1:01:01
How to Review Agent-Completed Tasks
Responding to a question about reviewing or testing tasks, the speaker says his main concerns are the solution’s structure, logic, output, and user experience. He notes that an application being declared complete does not make it truly finished, because proper review and pre-submission polishing may take longer than the initial build.
- 1:08:35
Project Update and Return to Streaming
The speaker directs late arrivals to the stream introduction for the new steps and ideas, says he is returning to streaming after a short absence, and considers making a separate project-update video.
Got a question about something in this stream?
Ask in the group