Meet the Swarm
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
A wide-ranging, practical stream about building AI products in public. Its strongest thread is the shift from short tasks to long-running “missions” pursued by agent swarms, supported by a control center and living roadmap that expose progress, quality, and blockers. Luqman discusses requirements documents and system design as foundations for greater agent autonomy, splitting work between Codex, Claude, or specialized models, and measuring real cost per completed task rather than advertised token prices. These are presented as concepts, early experiments, and personal judgments—not settled results. Practical activity includes reviewing an interface and identifying a scrolling bug, requesting an interactive mock to clarify the design direction, and beginning an application walkthrough by granting access to a sample project and requesting a quick capability test. The stream also discusses web testing and continuous monitoring, with the explicit caution that agent-driven UI checks do not prove scalability. Audience questions broaden the stream to RAG and learning resources, context compaction, usage limits and tool handoffs, handwriting OCR, reviewed data-entry automation, and disclosed AI assistants for social engagement. It also covers app frameworks and community platforms, product validation, free-plan pricing, distribution and sales, SEO, ads, direct outreach, freelancing, and uncertain effects of AI on jobs and SaaS. Open questions remain around Mission Control’s interface architecture, subagent concurrency, the value and cost of bridging models, and whether several proposed integrations and workflows will work in practice. The supported conclusion is that documentation quality, low-friction onboarding, human oversight, realistic usage measurement, and distribution matter at least as much as model capability; many demonstrated products and workflows remained exploratory or unfinished.
124 moments from the stream
- 0:52:05
Recheck Your Assumptions Before Blaming the Review Platform
The speaker advises a developer to review their own planning and research before attributing a setback directly to an app-review platform. Earlier investigation of review requirements, he argues, could have exposed the constraint and allowed an alternative path to be prepared.
- 0:53:24
Unexpected Project-Folder Duplication During Handoffs
While working with a coding agent, the speaker reports that each context handoff creates another complete project copy in the development directory and asks why this unexpected behavior is occurring.
- 0:54:08
When to Leave an AI Agent Room to Act
Responding to a question about rules files in Claude or Codex, the speaker says custom rules remain possible but argues that heavily constraining the agent may be less effective than leaving it sufficient room to operate.
- 0:57:29
Presenting a Personal Project as an Independent Developer
The discussion suggests presenting the owner of a personal project as an independent developer rather than a company.
- 1:03:01
Periods of Intense Work Focus
The speaker described a phase in which work feels more comfortable than socializing or travel as temporary in his opinion. He critically cited stories of extreme working hours and little sleep as unhealthy behavior. This was personal opinion, not a medical diagnosis.
- 1:04:56
Testing the Application with a Sample Project
The application walkthrough began by granting access to a sample project folder, requesting a quick test of its capabilities, and then asking for a full slide-deck theme mockup.
- 1:05:31
Comparing AI Tools for Motion Design
Responding to a question about motion design, the speaker said he had tried Claude with Remotion, Hyperframe, and other tools. He characterized the comparison as close, with different tools sometimes winning on particular tasks.
- 1:06:53
Connecting Tools and Agents in One Environment
The walkthrough showed a flow where a user selects a project and connects an agent. The speaker understood the concept as bringing external tools into one connected, closed environment instead of scattering them across separate places, and saw value in that approach for building systems.
- 1:09:12
Reducing User Setup Friction
The speaker observed that setup still required users to run commands and perform manual steps, and recommended making onboarding extremely easy. As a speculative idea rather than a tested result, he suggested a tool could let an agent such as Codex or Claude complete setup through computer use.
- 1:11:15
A Prediction of AI-Driven SaaS Commoditization
The speaker predicted that AI tools would lower the barrier to building SaaS products, potentially reducing their value as more people become able to create them. The discussion begins at the end of the page and remains incomplete.
- 1:12:49
From Individual Tasks to Coordinated Missions
The proposed application would still support ordinary tasks, but its stronger focus would be grouping related tasks into a larger “mission” and encouraging users to manage them as one coherent body of work.
- 1:13:52
Simplifying Connection and Onboarding
The feedback recommends learning from people who have tried similar ideas, preparing a project budget in advance, simplifying the interface and connection flow, and avoiding a setup that makes users run commands manually. It also stresses the growing importance of onboarding.
- 1:15:12
Could AI Reshape the SaaS Market?
Asked what comes after the current wave, the speaker says he does not know and avoids presenting a confident forecast. He relays the view that companies could benefit greatly if AI eventually performs functions now handled by external applications, while treating any decline of SaaS as an unresolved future possibility.
- 1:18:49
Building Tools on AI Platforms
The speaker discusses plugins and companion tools that operate within or support AI platforms. He offers the optimistic opinion—not a guarantee—that this direction gives new products a strong opportunity as AI takes a larger economic role.
- 1:19:58
Scoping a Command Center for Agents and Projects
The discussion begins by asking whether the tool is a CLI or IDE replacement, then frames it as a command center and project-management system that closely tracks each project’s goal. Software development is the first use case rather than the permanent scope; the longer-term vision includes research, content, and other work, with onboarding and appropriate access to tools and APIs required for coding sessions or agent swarms to operate autonomously.
- 1:28:22
Investigating an Issue Related to Facebook and Instagram
A partially described issue involving Facebook and Instagram is raised, followed by a recommendation to investigate the relevant tool or capability and use AI to learn more about it. The technology’s identity and the exact problem are unclear in the captions, so no confirmed diagnosis or fix is claimed.
- 1:32:32
Finding Similar Music While Respecting Copyright
Asked about the music being used, the speaker could not identify it precisely and suggested searching YouTube with a relevant keyword for similar tracks, while noting that copyright restricts some uses.
- 1:33:17
An Agent to Review Completed Work for Alignment
The discussion proposed running an agent after a major task is completed to review the work and assess whether it aligns with the intended result, using a handoff prompt to start the review. The speaker also expressed the opinion that higher-capability models may be more useful during ideation and creative work.
- 1:40:01
Store-Assistant Platform and Instagram Connection Walkthrough
The walkthrough ran into difficulty connecting Instagram and navigating the interface. In the speaker’s tentative interpretation, the tool appeared to be a small alternative or companion to e-commerce platforms, supporting website creation or management, replies to Instagram and Facebook customers, and order taking throughout the day.
- 1:43:16
Preparing Agents to Build a Product Autonomously
A proposed system begins with artifacts such as product requirements, entity-relationship diagrams, and a roadmap so agents can understand a product’s scope and execution needs. It also supports less-prepared users by helping them develop the documentation and system design needed before autonomous work begins.
Got a question about something in this stream?
Ask in the group