BREAKING MY APP TO FIX IT
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 follows an AI assistant for Shopify as development shifts from building to review, debugging, and UI/UX polish. Luqman evolves a command center that separates UI, UX, and backend verification from fix tasks, records outcomes, and supports step-by-step retesting, while routing work among Codex, Claude, and other agents according to the task. Practical work includes migrating the interface to Polaris, reviewing the homepage and report card, and diagnosing classifier, routing, conversation-button, image, live-counter, and relative-date issues. Some cases were verified as fixed, while others remained unresolved, including app slowness, conversation-flow problems, and a vision-analysis failure. Audience discussions cover model and coding-tool selection, voice-chat applications, and local versus cloud model access, with the comparisons treated as informal impressions rather than definitive benchmarks. On product strategy, Luqman argues for examining demand, unit economics, pricing, and customer-acquisition cost before scaling, while acknowledging that pre-validation is not the only path and that currency mismatch and purchasing power can complicate sustainability in some Arab markets.
105 moments from the stream
- 4:18:06
Enriching Fix Tasks with Context and Verification Steps
The discussion proposed pre-filling each task moved from review with its source, lane, security level, build/type-check/lint requirements, and concrete step-by-step operator verification. It also weighed a lightweight prompt per task against a full preparation phase using research and exploration agents, acknowledging that simplification may lose some enrichment.
- 4:29:49
Testing the Fix Workflow Incrementally
A small incorrect-date fix was chosen to exercise the new workflow, with a preference for gradual rollout instead of launching many agents at once. The speaker noted that debugging can be lighter than building, while some fixes will still require an actual build.
- 4:31:11
Exploring Alternative Models for Agents
An audience member reported that an unspecified model worked well and performed similarly to another model, prompting interest in setting up and trying alternative models later. This was an audience impression and a future intention, not a documented test result in the captions.
- 4:31:51
Separating Fixes from Verifications
A task was requested to be moved into the fixes section, while backend and new-flow tests were identified as verifications rather than fixes. Mixing the two categories could make the task board and workflow confusing.
- 4:32:38
Balancing Debugging Quality and Token Efficiency
Discusses a lightweight back-end prompting approach intended to preserve debugging quality while reducing token usage, with gradual testing before broader adoption.
- 4:36:04
Free Models: Local Deployment or Cloud Endpoints?
Explores the distinction between downloading models for local use and accessing them through free cloud endpoints, including questions about hardware requirements and a request for a link confirming whether free cloud access is available.
- 4:38:02
Separating Verification from Fix Tasks in a Debugging Workflow
Reorganizes the workflow so verification items remain in place, while a failed check generates a task that moves to the fixes section. UI polish, UX flows, and back-end items are then reviewed line by line to confirm correct classification before testing the flow.
- 4:44:16
Testing the Workflow with a “Last Month” Date Bug
A small bug is selected to test the workflow: a chat query for “last month” interpreted the period as December 2024 instead of March 2026. The example highlights the need to resolve relative dates against the correct temporal context.
- 4:47:00
Investigating Fixes When Work Starts
The session checks whether a minimal system prompt is behaving as intended. The explanation favors investigating a specific fix when work begins instead of analyzing every pending fix upfront and spending tokens on work that may not be attempted soon.
- 4:49:27
A Fix Followed by Interactive Debug Review
A debugging task is started without a separate preparation step. The intended flow is for the system to implement the fix, move it to review, and then guide the user through debugging step by step before completion; repeating this across many tasks is acknowledged as potentially heavy.
- 4:52:02
Fixing and Verifying Relative-Date Reasoning
A wrong-date response is attributed to the current runtime date not being supplied to the system prompt, with injection of that date proposed as the fix. Queries about the current month and two months earlier were used for verification, and the test returned the expected historical month in the stated April 2026 context.
- 4:52:51
Shopify App Load Time as a Separate Issue
The speaker identifies slow loading as a separate issue that may need fixing and states that Shopify rejects applications taking more than five seconds to load.
- 4:54:20
Debugging with the Model Intended for Production
Responding to a suggestion to use free endpoints, the speaker argues for debugging with the same models intended for the released product, because substituting another model—even under similar configuration—tests different behavior.
- 4:56:19
Allocating a Project Budget Across Priorities
The speaker recommends setting a budget in advance and deliberately allocating it among priorities such as advertising and debugging, rather than avoiding all spending without a plan.
- 4:57:17
Synchronizing Completed Tasks in the Fixes Board
Once a fix task is completed, the matching entry should automatically be struck through in the fixes section of the review file, keeping its status synchronized with the completed section of the task board.
- 4:59:14
Keeping Usage Limits from Blocking Testing
The question of which actions should be billable is deferred, while the free-task allowance is temporarily adjusted during pre-submission debugging so testing is not interrupted, with the intent to revisit it later.
- 5:00:56
Early Profitability and Break-Even Strategies
The speaker discusses the general goal of having project revenue exceed spending, while noting an alternative strategy: operate around break-even initially to acquire more customers, then raise prices later in pursuit of profit.
- 5:01:57
Starting Two Debugging Tasks in Parallel
After the single-task trial, two further issues—product-image rendering in chat and the task counter after a successful description rewrite—are started together to test whether the workflow can handle multiple tasks.
- 5:02:54
Validating Bug Fixes Step by Step
He describes reviewing debugging tasks one at a time and marking each complete only after opening the chat, asking a specific question, and checking for the expected response. Examples include a UI counter that updated only after a full-page refresh and a slightly distorted image while rendering a product page.
- 5:08:25
Restarting the Development Server and Checking Output
While checking whether the workflow is missing a step, he restarts the Shopify development server and then observes that the output file is still empty after startup. The excerpt does not establish the cause or a final resolution.
Got a question about something in this stream?
Ask in the group