Talkstore Begins Now! 🟢 Building to $100K
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 the public start of building Talkstore: a proposed Shopify-specific AI assistant through which merchants could manage products, storefront work, and commerce intelligence conversationally. Its capabilities and possible expansion beyond Shopify are presented as conditional directions, not validated outcomes. Practical work covered Shopify app and development-store setup, Supabase integration and Prisma migrations, secret handling, and repeated diagnosis of session, connection, preview, and Railway deployment problems. The preview and some setup steps worked, and the service later appeared active, but several database, network, build, and deployment issues remained intermittent or unresolved in the covered segments. This led to useful lessons about starting with logs, documenting a stubborn problem, seeking a fresh diagnostic perspective, and reviewing coding-agent compromises rather than accepting a result merely because it works temporarily. The stream also presents a multi-agent command center that separates research and task preparation from implementation, code, security, and compliance review, using dependencies, an adaptive timeline, and human approval gates. Luqman felt this structure accelerated and preserved the context of the work, while still questioning whether every agent was running and visibly logged. He emphasized that authentication, secrets, and nuanced decisions continue to require human intervention, and that task count is a poor measure of difficulty. Work later shifted toward pre-launch distribution: a simple but polished Astro waitlist site kept separate from the Shopify app, with email capture, analytics, social metadata, a teaser, and a distribution campaign. The first landing-page draft had missing hero content and styling or rendering failures, so it was treated as an unfinished iteration. Other useful discussions included audience questions about using AI to explore income paths from existing expertise, SaaS unit economics, choosing between ads and SEO/GEO, Arabic-market WhatsApp automation, and targeting external markets, plus local voice dictation, content automation, and Telegram community options. Stream lag and interruptions were also investigated without a conclusive cause, while music licensing and gym culture appeared as brief secondary discussions.
121 moments from the stream
- 3:24:36
Troubleshooting Deployment and Supabase Environment Variables
While a deployment was failing, Luqman reviewed the database URL and service-level environment variables, noting that Supabase provides different connection URL types. He also explained why configuration screens containing keys and secrets should not appear in a public replay: exposed credentials could enable unauthorized access or changes. The page does not confirm that the deployment ultimately succeeded.
- 3:30:15
Checking SaaS Unit Economics Before Building
Luqman advised checking margins, monetization, and profitability before building a SaaS product, especially when API costs are denominated in dollars while customers pay in weaker local currencies. In his view, this can constrain scalability in markets such as Egypt, Morocco, and Algeria, though he qualified it as market-dependent. He suggested discussing unit economics with an AI assistant early rather than discovering an unprofitable model after building.
- 3:31:47
Why Task Count Does Not Reflect Development Difficulty
Luqman described a slow debugging loop: wait for a deployment, receive an error, review it with coding agents, and deploy again. He contrasted this with his usually simpler experience on Vercel and Cloudflare, then moved to a Railway pipeline task. His broader point was that a week with fewer tasks can take longer than one with more tasks because a single blocked task may consume substantial time.
- 3:35:48
Debugging a Failed Deployment and Setting Up Railway Auto-Deploys
He reviewed Railway logs after repeated service deployment failures, tried removing the host variable, and redeployed. Once the service appeared active, he described a pipeline using a corrected Dockerfile and automatic builds when changes are pushed to the branch.
- 3:44:17
Recording the Command-Center Workflow
He explains that a JSON tracker records activity in the command center, supporting sequential handling of sections and tasks and a return to a single terminal session.
- 3:47:10
An Opinion on Targeting External Markets
In a brief audience exchange, he shares the opinion that building a business for markets outside one’s home country may offer better opportunities, framing it as encouragement rather than a guaranteed outcome.
- 3:47:51
From Task Preparation to Build and Review
He describes a multi-agent workflow: the system reads the task definition, deploys exploration, research, and project-management agents, inspects the codebase and Shopify documentation, and prepares an instruction file for a builder agent. After the task starts, the builder applies it and another agent reviews the code for problems; he notes that the agent roles still need clearer documentation.
- 3:49:48
When a Task Needs Collaboration Mode
He defines collaboration mode as a case where a human operator works with the agent, and presents task preparation as an investigation and research phase that produces instructions for a later builder agent.
- 3:54:18
Talkstore’s Initial Shopify App Skeleton
He showed the still mostly empty app skeleton inside Shopify and said a follow-up check indicated it was working. He noted that his prior experience was mainly on the merchant side and with the Shopify ecosystem rather than app development itself.
- 3:55:50
A Vision for a Command Center That Turns Project Context into Plans, Agents, and Tasks
He said the current command center and swimlane board were helping him move faster. He then outlined a prospective downloadable tool that could inspect a project’s context, configure a swimlane, generate nodes, tasks, and agents, and estimate the work; he was still testing whether the approach genuinely improved productivity and had not decided whether to release it as a standalone, open-source, or commercial tool.
- 4:00:03
An Idea for Focused Guides on Building and Marketing AI Products
He conditionally proposed short, focused courses covering AI and agent tools, current information, business building and marketing, and monetization design. He tied the idea to sufficient audience interest and did not present it as a confirmed plan.
- 4:01:41
Building in Public Over Chasing Views
He said he did not want to focus heavily on asking for likes and subscriptions or chasing views, partly because the product’s target customers differed from much of the channel’s regional audience. He added that sufficient interest could still create substantial value from watching someone build the product live in public.
- 4:02:46
A Multi-Agent Workflow from Task Research to Implementation and Review
He reviewed a workflow that prepares a task, assigns agents to inspect the codebase, research it, and produce an implementation-ready prompt. Once started, a builder agent applies the work, after which it automatically passes to later agents including security and review roles.
- 4:04:41
Moving a Telegram Presence from Broadcast to Community Discussion
The discussion clarified that a Telegram channel is primarily for broadcasting, while a group allows everyone to participate, and that a public link does not make a channel conversational. The speaker explored discussion settings and linking a group, then decided to research and make the change later while retaining existing members if possible.
- 4:10:33
Protecting App Keys and Secrets During Live Setup
The speaker hid the screen during app setup because the steps involved encryption keys, API keys, secrets, and environment variables. He noted that this kind of setup and minor troubleshooting is common in the first tasks before moving on to subsequent work.
- 4:14:36
Choosing Background Music with Copyright in Mind
The speaker discussed the difficulty of choosing background music that supports focus without causing copyright issues, noting that no-copyright music may be found or purchased. Panda Beats was mentioned as a suggested source, without any documented test of its content or licensing terms.
- 4:26:06
Diagnosing a Prisma Session-Table Migration Issue in a Shopify App
The discussion starts by clarifying which Shopify development store to open, then traces an error stating that the Prisma session table does not exist. The work includes running a Prisma migration, restarting the Shopify app development environment, and reviewing migration status and general Supabase connection configuration. The app appeared to load briefly, but later output remained incomplete and the investigation had no confirmed final resolution.
- 4:28:40
Unexpected Automatic Compaction During Long Streams
The speaker recalls that a previous seven-hour stream underwent an unexplained automatic compaction or shortening after roughly five and a half hours and says he wants to investigate its source. No cause or responsible system is established.
- 4:34:26
Troubleshooting Supabase Integration with a Shopify App Installation Flow
A practical attempt to connect a Shopify app’s authentication and installation APIs to Supabase. The development environment refused to connect and kept loading, prompting inspection of the error and a stop-and-restart attempt before continuing the flow.
- 4:40:41
Verifying Access-Token Encryption for Shopify Requirements
The access-token column was checked during work associated with Shopify’s app-submission checklist, which was described as requiring token encryption. After refreshing the development store and reviewing Supabase, the token appeared as a Base64 string and the encryption check was treated as complete.
Got a question about something in this stream?
Ask in the group