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
- 2:00:23
Checking future posts and changing the scheduling interval
Requests checking the calendar for pending posts on future dates, then changing the scheduling interval from every ten minutes to every three hours. The captions provide no confirmation of execution or check results.
- 2:01:25
Why separating agents does not guarantee parallel work
Luqman describes splitting short-form production, videos, and audience replies among specialized agents to reduce workload. Agents still wait on one another, exposing a coordination problem for which no implemented solution is presented here.
- 2:03:31
Exploring 3D animation and modeling tools
The discussion explores a 3D animation tool, user experiences, and cheaper alternatives. It includes a reported image-to-3D example and mentions free trials and offers available at the time of recording, without establishing a completed trial by Luqman.
- 2:06:03
AI-assisted application review and task handoff
Luqman reports reviewing an application with a model and fixing issues found during that review. He then proposes handing a task to another model with on-demand oversight and asks for the identified problems to be explained before proceeding; independent verification tests are not documented.
- 2:12:56
Assessing an automation idea requires a clear example
Luqman declines to assess an idea he does not yet understand and requests a reference to clarify it. He offers a tentative positive outlook on agentic automation.
- 2:14:26
A voice assistant concept that turns goals into delegated work
Luqman outlines a unified voice platform with a hoopoe persona that would break user goals into tasks and delegate them to specialized agents. He frames its intended role as a manager or work partner beyond orchestration, while describing its foundations and agent harness engineering as work in progress.
- 2:17:37
Choosing an automation niche around anticipated needs
Asked which automation niche to pursue, Luqman personally recommends building things that support AI's development. He suggests researching anticipated needs over the following six months and into 2027, then building tools around them, presenting this as opinion and forecasting.
- 2:19:59
Reviewing the hoopoe character's eyes
Luqman identifies the eyes as an unsatisfactory part of the hoopoe character design that needs refinement.
- 2:24:09
Choosing a niche for AI products
Luqman discusses the difficulty of finding an innovative idea and suggests building for an AI direction expected to grow over the coming months. He uses AI coding tools as an example of serving a specific niche within a broad field.
- 2:25:34
Envisioning voice interfaces and goal-driven work
Luqman speculates that current coding editors and interfaces could give way to a configured platform where users express their goals by voice. He connects this with less manual prompting and continuously running agents, while acknowledging uncertainty and proposing an experiment.
- 2:29:56
Reducing human intervention as AI capabilities grow
Luqman expects improving AI capabilities to gradually reduce human prompting and enable broader automation. He says he is trying to build something for that direction, presenting it as an expectation rather than a demonstrated achievement.
- 2:37:06
Focusing on one product and estimating investor equity
The speaker recounts a founder closing side projects to focus on one product and expresses optimism about that decision. He then discusses company valuation and investor equity, explicitly treating the proposed percentage as a rough, uncalculated estimate.
- 2:40:50
Requesting a diagnosis of recording-processing slowdowns
A request to inspect the workflow for processing a recording backlog asks whether the workload explains the computer slowdown. These captions provide no confirmed diagnosis or repair outcome.
- 2:45:59
Meshy versus Tripo: pricing, textures, and game prototypes
The discussion compares entry prices displayed at the time of recording while leaving credit costs unresolved. An audience contributor reports preferring Tripo's textures after free trials of both tools, a judgment limited to those trials. The discussion also covers trying 3D tools for game prototypes dating to 2019.
- 2:51:09
Implementation delays and a perception of Claude's speed
The speaker describes frustration with work progressing more slowly than his ideas and says Claude appears faster. This remains a personal impression; no measured comparison is presented in these captions.
- 2:52:23
Game vibe coding and animation tools
The discussion describes a creator building a game through vibe coding and mentions Blender in an animation workflow. Some tool names and the sequence of their use are unclear in the captions.
- 2:53:14
Reassessing whether a video background is real
Luqman initially doubts a video background because the clouds appear stationary, then retracts that impression after noticing movement and concludes it is real. This is his assessment during the discussion.
- 2:54:28
Marketing apps to a creator’s audience
Luqman finds a creator’s frequent switching between AI tools confusing and criticizes lengthy motivational talk and app promotion. He discusses building products to sell to an existing audience while praising another creator’s approach.
- 2:59:25
Matching eyelashes to a generated reference
An English-language exchange requests eyelashes matching the originally generated reference. The captions do not establish the tool being used or whether the change was completed.
- 3:03:49
Planning a graph-based automation system with mission control
Luqman outlines an evolving idea combining graph-based automation with mission control. Work includes the website, core engine and specifications; he calls agent harness engineering a double-edged sword without detailing the tradeoffs here.
Got a question about something in this stream?
Ask in the group