Four Projects. One Flow.
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
Luqman and the audience explore a platform that would turn conversational or spoken goals into teams of agents for building products and running operations. Topics include roadmaps, role contracts, architectural oversight, and knowledge and context graphs; the concept remains under development and its feasibility unresolved. Token consumption, repeated file reading, review loops, and blocked parallel work recur throughout, alongside proposals for event-triggered supervision without a practical comparison establishing its advantage. Practical work includes reviewing an animated hoopoe website, exploring 3D and reverse-engineering tools, and demonstrating keyword search in a partially indexed stream library, with semantic search proposed. Content automation, reply approvals, and scheduling are also reviewed, including posting that reportedly continued after a pause request and unresolved performance and storage problems. Audience questions cover learning AI, choosing a specialization, and reaching customers; Luqman recommends planning distribution and using direct outreach for B2B. His predictions about voice interfaces and increasingly autonomous agents remain forecasts, while the work presented does not establish a completed platform or successful end-to-end automation.
153 moments from the stream
- 1:04:10
Copywriting quality can differ from coding quality
The speaker relays an audience observation that some models, including Sonnet, produce good code but weak project copy, and broadly agrees. These are discussion opinions and experiences, rather than systematic comparison results.
- 1:06:29
Scheduled-task token use depends on the workload
The speaker discusses how a scheduled task’s token use depends on its work, using the example of checking an agent’s status and restarting it when idle. He suggests a simple check may be lightweight, without establishing a general measurement or claiming that all scheduled tasks are inexpensive.
- 1:09:36
Progress reporting and visual oversight for large agent tasks
The speaker requests percentage-based reporting of completed and remaining work and expresses reservations about assigning an entire large milestone to an agent without execution-management tools. He proposes visual oversight of agents and sessions to spot potentially wasteful token use; the page does not establish a completed implementation.
- 1:10:48
Agent harnesses: loop detection and customization risks
Luqman describes a rule that flags repeated agent tasks or loops as stuck, enabling intervention to reduce wasted tokens. He cautions that custom harnesses can underperform those provided by Claude and Codex, and argues that software engineering experience helps with this work.
- 1:17:28
Claude and Codex: usage resets and workflow preferences
Luqman notes differences in usage resets between Claude and Codex. He expresses a preference for continuing to use Claude while describing Codex as capable of handling a broad range of tasks.
- 1:19:12
Searching a stream library for relevant moments
Luqman demonstrates searching a stream library for Codex, explaining how results connect recordings to relevant moments, including audience questions. He says the interface is still experimental and plans to improve it.
- 1:23:10
Severe lag obstructs application use
The speaker reports application lag severe enough to obstruct use, without identifying a cause or presenting a fix in this passage.
- 1:24:30
Featuring AI employee characters on a teaser website
Proposes featuring AI employee characters with distinct faces on a teaser website, alongside a group of birds with one main bird for interaction. These are proposed ideas, with no implementation established.
- 1:27:54
Debugging and polishing ahead of anticipated installs
Reports debugging and polishing the application internally ahead of an anticipated increase in installs in the last week of October. Explains that it offers a free plan rather than a free trial, and that reviewing it uncovered bugs.
- 1:30:08
Pricing changes and visibility against competitors
Notes changes to the application's pricing and argues that greater traffic and visibility help explain competitors' advantage. A further competitor comparison begins but remains incomplete on this page.
- 1:32:17
Traffic before judging pricing
The speaker argues that a strong product still needs visibility, and that traffic helps reveal product performance and inform pricing decisions. No marketing campaign outcome is established.
- 1:33:15
A 3D bird that responds to scrolling
A proposed animated bird would briefly disappear and return as visitors scroll. Creating the 3D bird is identified as the first step; no completed implementation is shown in the captions.
- 1:33:57
Collaborative contributions to product development
An idea is raised to bring contributors together to build a product, without specifying a collaboration mechanism or confidently recalling its name.
- 1:35:32
Automated posting continues after a pause request
Automated posting is reported to have continued despite a pause request, prompting an urgent request for a fix. The page does not establish the cause or confirm a repair.
- 1:37:00
Application lag with an uncertain cause
Lag is noticed during use, with uncertainty over whether the application or the computer is responsible. No diagnosis is established.
- 1:38:24
Requesting a storage usage breakdown
A storage usage breakdown is requested to understand space consumption and the need for more capacity. No completed analysis or cleanup is presented.
- 1:39:43
Reviewing social media reply workflows
A workflow configuration review is requested to understand replies on TikTok, Instagram, and direct messages. The page provides neither the actual configuration nor review findings.
- 1:41:18
Claude permissions and Mac storage and performance issues
A reference to reviewing Claude’s removable-volume permission is followed by concerns about Mac storage and performance. The captions establish neither the cause nor a resolution.
- 1:42:44
Controlling and approving content-agent replies
Questions address reply triggers, how often to mention a point, and limits on repeated comment exchanges. Luqman describes bulk approval with editing available and raises excessive posting as another concern; implementation of a reply limit is not established.
- 1:59:26
Clarifying a question about intelligent automation
An audience question is clarified as describing a system that can handle unexpected events. Luqman requests a company example or reference to understand the intended field before offering a view; these pages contain no detailed assessment.
Got a question about something in this stream?
Ask in the group