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
- 3:12:01
Before Building a SaaS: Plan Monetization and Distribution
In response to a question about launching a SaaS in the Arab region, the advice is to begin with monetization and distribution strategy: market demand alone is insufficient without a realistic way to attract traffic and sell the product. A detailed competitive teardown can reveal where competitors get traffic, while AI suggestions should be checked against practical research.
- 3:14:43
Content Automation Does Not Guarantee Reach or Traction
The speaker discusses using groups of agents to produce posts and messages and distribute content across multiple platforms, but explains that output volume alone did not deliver the desired result. The practical lesson is that traffic acquisition remains the core challenge and may require less reliance on inbound content and more emphasis on outbound outreach or advertising.
- 3:17:47
Inbound vs. Outbound Traffic
Inbound marketing is compared to a magnet: a business publishes content or advertisements that draw people toward it. Outbound marketing means approaching the target customer directly. Joining a trend that attracts new visitors is given as an inbound example, while ads use data signals for targeting and outbound outreach can focus on a defined customer persona.
- 3:19:58
Inbound Ads and Research-Driven Outbound
The speaker describes marketing and traffic acquisition as difficult product stages, distinguishing inbound ads from outbound outreach that starts with researching people who match the target persona and then contacting them.
- 3:21:35
A Searchable Knowledge Base for Stream Content
The speaker describes SHA8AL.com as a community site for showcasing members’ projects and building a knowledge base from stream transcripts, where a search for a tool or term can lead to the relevant stream and location. Indexing the archive is presented as a substantial retrospective task before processing new streams.
- 3:25:45
Conversion Signals and Ad Campaign Stability
Based on the speaker’s experience, ad platforms need enough weekly conversions to improve their data signals and targeting; he cites roughly 50 conversions as a minimum rule of thumb. He suggests choosing a more attainable objective for the budget or increasing the budget, while qualifying that campaign stability is only relative and that installs for a free product may cost differently.
- 3:29:11
When Do You Need AI Agent Swarms?
Answering a question about a graduation project that simulates a company with multiple agents, the speaker says agent swarms are not automatically appropriate for every project. The choice depends on the tasks, and a single agent may sometimes coordinate sub-agents; his core recommendation is to keep the design simple.
- 3:31:33
Once the Product Is Ready, Focus on Traffic
After praising the design of a project shown to him, the speaker identifies traffic acquisition as the main remaining constraint and advises concentrating effort there, especially once the product itself is complete.
- 3:35:43
Community and Webinars as Alternatives to Direct Ads
The discussion explores attracting traffic by building a community around an audience need and promoting an educational online event, such as a sales-focused webinar, instead of relying only on direct ads for the tool.
- 3:36:59
Sarmad as an Automation Graphing System
The speaker describes Sarmad as an automation graphing system, clarifying that calling it a “command center” does not precisely reflect its purpose.
- 3:37:46
Estimating Acquisition Cost and Testing Ad Channels
A discussion of using available results to estimate cost per install and budget needs, while testing multiple advertising channels instead of relying on one platform. Private targets and actual figures are excluded.
- 3:39:07
What a Graph Tool Needs Before Public Release
The speaker explains that publishing the graph tool on GitHub involves more than creating a repository: its structure must support uses beyond his own, with clearer organization and an onboarding experience. The timing remains uncertain.
- 3:44:12
A Conditional Assessment of an AI-Employee Coordination Project
An assessment of a graduation project with substantial conceptual overlap with Sarmad. Its strongest aspects are the architecture and the potential for an evaluated security, coordination, and controlled-delegation system; its main risk is excessive scope before proving that the workflow truly benefits from AI. A GPT-assisted analysis presented during the session conditionally endorsed the direction as a graduation project.
- 3:46:32
Customer Acquisition Beyond Ads
A brief answer on alternatives to advertising, including outbound and inbound acquisition flows, automation, and direct email outreach to prospective customers.
- 3:49:40
AI and Jobs: Displacement Versus New Roles
The speaker resists framing AI as simply good or evil, comparing it with the industrial and internet revolutions, which displaced jobs while creating others. He emphasizes saved time and effort and, in his view, the continuing importance of a human in the loop.
- 3:51:09
Paperclip as a Hierarchical AI-Agent Platform
The speaker describes Paperclip as a platform that brings hierarchically organized agents into one place and allows different agents to use different models. He characterizes it as a graph system within a specialized platform, tentatively describes it as open source, and notes that it is on GitHub.
- 3:52:56
Organization Pages as an Organic Search Strategy
A participant describes an organic-discovery approach that creates a dedicated page for every organization listed on a platform, allowing that page to appear when someone searches for the organization on Google. The idea was discussed in an SEO and blogging context, without results being independently verified during the session.
- 3:54:58
From Prompt Engineering to Goal-Directed Agents
The speaker said the project was relevant to AI work at the time of the recording, describing a historical progression from prompt engineering in 2023, to loop engineering in “the previous year” as he put it, and then graph engineering. As a future direction, he proposes giving agents a shared goal instead of detailed tasks and prompts, allowing them to create subordinate agents to pursue it, while explicitly noting that this is a future stage rather than a capability already achieved.
- 3:57:02
Simulating How an AI Company Operates
A proposal to build an environment that represents or simulates how an AI company operates.
- 3:57:22
Using Organization Pages to Improve SEO
A discussion of legally using client data and creating organization-specific pages to attract clicks and improve search ranking; private operating figures are omitted.
Got a question about something in this stream?
Ask in the group