New SHA8AL.com + Sarmad
Stream summary 75This stream in minutesWant 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 centered on turning SHA8AL.com into a community hub and searchable bilingual knowledge base that links topics and tools to timestamps across the broadcast archive while showcasing members’ projects. Luqman discussed an automated workflow for downloading YouTube captions, filling gaps locally, and then using models to summarize, select passages, and produce short and longer videos. Arabic transcription accuracy, selection criteria, video quality, and archive-processing cost still required evaluation, and no completed system was demonstrated. Practical work included reviewing Burnish and smoke-testing its installation and free-plan flow, with proposals to reduce onboarding friction, clarify what counts as an install, improve the analytics dashboard, and track Shopify App Store visits. A recurring conclusion was that distribution and traffic—not product construction or the funnel alone—were the main bottleneck, prompting exploration of measurable channels such as outbound outreach, agency partnerships, advertising, and communities, without claiming final results. The stream also covered Sarmad as an automation graph system, Paperclip and goal-directed agent coordination, build-versus-buy tradeoffs, and privacy and usage-limit constraints. Luqman offered qualified, non-benchmark impressions of Codex, Claude, GPT, transcription, and video tools, emphasizing hands-on testing, adequate context, and validation when model outputs conflict. Audience discussions additionally touched on building a Micro-SaaS with AI, market choice, freelance pricing, AI’s impact on jobs, and automated social replies, while several proposed workflows, comparisons, and tests remained unresolved.
177 moments from the stream
- 6:16:29
When AI Models Contradict Each Other on Growth Strategy
The speaker describes giving two models the same prompt to produce a traffic strategy, then asking each to judge both proposals. Their verdicts conflicted sharply, with one model criticizing its own proposal, leading him to conclude that these outputs alone were not a reliable basis for a marketing decision.
- 6:22:05
Comparing AI Tool Costs Through Real Usage
In a discussion of API and subscription pricing, the speaker says theoretical calculations are insufficient and that a better comparison comes from hands-on use and understanding the usable volume and token allocation each plan provides.
- 6:23:16
Measuring model cost by tokens consumed per task
The speaker argues that tokens consumed per task can be more informative than the price per million tokens, since a nominally cheaper model may use more tokens to finish the work. Some model names in the example are unclear in the captions.
- 6:24:33
What is traffic? Inbound and outbound acquisition explained
After stating his opinion that AI cannot guarantee traffic, the speaker explains it through a shop-in-a-mall analogy: traffic is the people entering the shop, outbound acquisition means actively calling them in, and inbound acquisition relies on attractions that draw them inside.
- 6:28:22
A subjective comparison of Codex and Claude
Participants describe Codex and Claude as closely matched, while reporting different personal preferences and experiences. The discussion concludes that the choice is subjective when tools perform at similar levels.
- 6:31:23
Improving long-video transcription and automatic clip extraction
A gaming creator asks how to extract important clips from a long video when the Arabic transcript contains errors. The discussion stresses that transcription accuracy is essential for selecting publishable topics and clips, suggesting repeated transcript review and correction or a better transcription method, while noting that extra passes consume more tokens. The workflow can then produce timestamped topic ranges and become substantially automated.
- 6:32:14
Price versus quality when choosing AI tools
One participant offers the qualified opinion that some users and companies now prioritize price over differences in quality, while acknowledging that this varies by situation.
- 6:35:24
A dated opinion and prediction about changing AI pricing
The speaker criticizes subscription-tier changes he says occurred after October 28 of the recording year, then predicts—as speculation rather than established fact—that pricing and efficiency could change after a potential IPO.
- 6:41:29
An Automated Workflow for Stream Transcription
A workflow places each completed stream recording in a watched folder, where a scheduled script detects it and hands it to an agent for transcription and subsequent processing.
- 6:44:47
Understanding a Transcription Script with Codex
The discussion considers a Google Colab script instead of relying on CapCut solely for transcription, and suggests giving the script link to Codex to explain what it does and how to try it.
- 6:50:48
Should a Reply Bot Disclose That It Is Automated?
An audience proposal suggests an agent that learns a reply style from human examples before handling social comments and messages, raising the question of whether openly identifying it as a bot could feel more authentic and attract attention.
- 6:52:41
Conversion Matters More Than Traffic Alone
Responding to a suggestion to use the community for traffic, the discussion notes that only part of a community is typically active and argues that useful conversion matters more than traffic volume alone.
- 6:53:35
Designing a Reply Bot with a Distinct Personality
The speaker argues that people may ignore bots associated with rigid canned replies, and instead explores a clear, playful personality that responds accurately and helps users; training it on his own style remains a possibility rather than a final decision.
- 6:56:28
Deriving Video Ideas from Multiple Transcripts
A proposed bot would combine transcripts from several videos and use a prompt to extract strong ideas for new video content. The response said that, at the time of the recording, they were already applying a similar idea, and the speaker believed many others used comparable workflows.
- 7:00:12
An Agent for Social Comments and DMs
A discussion about creating an agent with a defined personality to reply to comments and direct messages on Instagram, TikTok, and YouTube, including whether to build a new agent inside Sarmad and how the workflow could be automated. No completed implementation or test is shown.
- 7:00:49
Progressive Techno as Focus Music
A personal opinion that background progressive techno, especially through headphones, can support concentration while working and fade from conscious attention over time. This is an anecdotal view, not a tested finding.
- 7:03:38
Checking Manifesto and Pitch Deck Alignment
A request to review the newer pitch deck as the current expression of the vision and mission, then compare it with the earlier manifesto to identify any misalignment or determine whether the manifesto was updated when the deck was created.
Got a question about something in this stream?
Ask in the group