26 AI Employees → 9 Stronger Ones
Stream summary 74This 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 documents broad practical work and discussion around building, marketing, and measuring AI-assisted products. Its strongest working thread was a Shopify app: reconsidering free-plan versus trial pricing, reviewing onboarding, code, and testing, and developing GA4 analytics that connect visibility, traffic, and orders. Deployment and test outcomes and the pricing impact were not settled, while Luqman viewed acquiring enough traffic as the main remaining challenge. Growth discussions covered inbound and outbound channels, demos, partnerships, advertising, separating traffic problems from conversion problems, demand validation and distribution, and possible later expansion into more languages, platforms, and markets such as Salla. They also included general opinions on bootstrapping, investment, when to raise funding, avoiding excessive early equity dilution, the trust barrier in government sales, and market-dependent YouTube monetization. The stream also explored smaller AI-agent teams with defined roles, dependencies, and human oversight; a permission-limited central executive agent; and voice-assistant tests that exposed latency, interrupted conversations, conflicting records, and incomplete integrations. Luqman compared coding tools and models including Codex, Claude, Gemini, and Grok, arguing that token prices and published benchmarks are insufficient without task-specific efficiency measurements and consistently framing his assessments as personal observations. Codex slowdowns and context degradation remained unresolved. The attempt to locate Gemini 4 was specifically a historical snapshot dated October 1, 2026 and did not establish its availability or final identity. Further discussion covered goal engineering, data and website security and AI’s use by both attackers and defenders, hardware constraints on local models alongside the opinion that local compute may matter long term, practical AI learning while continuing computer-science study, and whether AI weakens thinking or improves efficiency. Other topics included remote work and the possible advantage of lower-cost locations; general distinctions among work or study migration, asylum, and visits; non-personal opinions about financial independence, investment funds, and dollar-cost averaging; and workspace ergonomics through standing desks or walking pads. Luqman also discussed content and livestream strategy, including consistency versus repetition, streaming for accountability, useful standalone videos, and background-music copyright risk. The music work went beyond a general focus-music experiment: it covered tempo, layering, and Arabic maqamat, plus a Suno Studio workflow in which a musical scale is created with Python, uploaded, arranged in layers, and supplemented with sounds. This work and the assistant’s visual and vocal identity remained experiments rather than demonstrated successes.
180 moments from the stream
- 5:24:59
Reviewing GitHub Projects and Skills with AI
A participant described bringing a GitHub project into an AI tool for inspection and agent generation, while Luqman explained that he similarly asks ChatGPT whether a GitHub skill would help his project. He had not used the other service, so he could not confirm its usage limits and only speculated that it might be cheaper.
- 5:26:48
Assessing E2E Test Results and the Need for an Adversarial Session
Luqman checked the work status by asking whether an E2E smoke test had been performed, whether critical issues remained, and whether a separate adversarial session was needed. The response is unclear in the captions, so the test outcome and fixes cannot be confirmed.
- 5:28:24
One Personal Assistant or a Team of Agents?
Based on his current understanding, Luqman compared a system offering one personal-assistant agent with built-in tools and scheduling against a system that can create multiple agents as a working team. He qualified this as a view of the current products that could change, and said the single assistant's main distinction from a regularly scheduled session appeared to be its preconfigured setup.
- 5:32:11
The Complexity of Configuring GA4 Through Google Cloud
Luqman described GA4 as a specialized, complex platform with dedicated courses and specialists, while the agent was performing configuration without directly opening its interface and working through Google Cloud behind the scenes. A comparison with Meta's platform begins at the end but is incomplete on this page.
- 5:35:25
Deep Code Audits with an Adversarial Lens
An explanation of a deep-audit approach that distributes reviews of code, pull requests, and tests across multiple agents, asking them to assume something is wrong until the system proves otherwise. The speaker noted that this adversarial stance may introduce bias, but can also uncover issues missed by ordinary review.
- 5:39:10
Why Decisions Need Analytics Beyond Platform Reports
A discussion of the limits of built-in platform analytics and the value of adding tools such as Google Analytics 4 to obtain a clearer view of data and make deeper decisions. The speaker explicitly remained uncertain about mobile app-store analytics.
- 5:40:15
The Role of an AI Agent in Trading Research
A general discussion of the difficulty of scalping and the idea of using an AI agent to prepare reports and identify possible entry opportunities, helping a human make a more data-informed decision rather than automatically executing trades. The field was characterized as difficult and deep.
- 5:41:05
Does AI Weaken Our Ability to Think?
Responding to whether AI weakens thinking, the speaker suggested that the risk may arise when a system thinks on the user’s behalf, not merely when it performs tasks such as coding or marketing. He compared the debate with calculators and questioned how “weaker thinking” could be measured, without reaching a definitive conclusion.
- 5:43:21
Can Another Analytics Tool Replace PostHog?
A brief opinion that PostHog is useful with many users, alongside consideration of replacing it with another analytics tool; the alternative’s name is not fully clear in the captions.
- 5:43:49
Does AI Make Us Less Skilled or More Efficient?
A discussion comparing AI’s effect to the calculator: it may reduce some manual effort while raising quality expectations and saving time on repetitive tasks. Whether that is laziness or smarter time management remains a matter of opinion.
- 5:45:16
Advice for a Computer Science Student in the AI Shift
The advice is to continue studying, develop a specific skill with AI’s help, and follow developments without becoming overwhelmed by the field’s rapid changes.
- 5:48:55
Defining Activation Across Free and Paid Plans
A think-aloud discussion about whether a free-plan install should count as activation, and whether a separate metric for paying users is needed before locking in the measurement definition.
- 5:56:31
Why Token Price Alone Cannot Compare Model Costs
The speaker explains that price per million tokens does not by itself determine task cost: model efficiency and the volume of input and output tokens consumed also affect the total. He presents this as his interpretation of why models with different headline prices can sometimes have similar practical costs, while some model names and figures in the captions remain uncertain.
- 6:01:48
A Brief View on Investment Funds and DCA
As a general opinion rather than a guaranteed recommendation, the speaker suggests that investment funds may reduce the need to manage individual stock trades personally, and mentions dollar-cost averaging (DCA) as a core concept to understand. The excerpt does not address risks, fees, or differences among fund types.
- 6:04:43
Analytics as the Project’s Lens
The speaker describes analytics and data points as a lens for seeing traffic sources, user paths, and conversions, arguing that business decisions require clear measurement.
- 6:07:02
Consolidating Measurement Sources in Google Analytics 4
The speaker describes ongoing work to consolidate data from several platforms into Google Analytics 4 as a central measurement view, connecting advertising, organic-traffic, and ecommerce sources.
- 6:08:33
Using AI to Configure Tracking
The speaker argues that AI can help configure Google Analytics 4 tasks such as filters, properties, and data streams from a description of the desired tracking events. This is presented as his opinion, not a documented test result in the excerpt.
- 6:10:43
Testing an AI Assistant Connected to Saved Records
The speaker tests an assistant that answers from saved records, and the assistant clarifies that its view is a stored snapshot rather than a live display. The subsequent feedback focuses on response latency, making replies more concise, and tuning noisy notifications.
- 6:13:37
Speed Versus Cost in Live Interaction
The speaker says he accepts a higher cost when fast live interaction is needed, arguing that greater execution efficiency may matter more than simply choosing the cheapest service.
- 6:16:50
GPT Slowness Even on a Documentation Repository
The speaker says the slowdown appears general rather than caused by the current repository, which is a separate documentation-only repository. He mentions seeing a similar report on X but says he does not know the cause; an audience suggestion that it was intended to encourage payment for Fast Mode remained speculation.
Got a question about something in this stream?
Ask in the group