Ads, Data & AI Marketing Agents
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
The stream follows the connected work of building and marketing a Shopify app: testing it in a real store environment, seeking additional perspectives and user feedback, monitoring speed and bugs, and making the product’s value quickly understandable. Much of the discussion concerns advertising readiness and measurement—reviewing creatives, correcting social-page branding, consolidating platform metrics into a data pipeline, and choosing a conversion event that balances signal volume with proximity to the real goal, such as installation or completed onboarding. Luqman argues that creative intuition alone is unreliable and that testing, market response, and data matter more, while stressing that trials need time to mature and that the campaigns had not yet demonstrated success. The group also discusses Meta’s shift toward broad targeting, “learning limited” campaigns, lead-capture funnels, and attribution, leaving questions such as the best optimization event, the value of lookalike audiences and affiliate marketing, and parts of tracking and implementation unresolved. The stream also presents experiments and proposals involving Claude, Codex, MCP, and specialized agents for campaign setup, ad review, cross-session strategy handoffs, content production, SEO, and social publishing. Luqman emphasizes human review because AI recommendations can become inconsistent, and describes designing an agent that challenges assumptions instead of automatically agreeing. Audience questions cover possible Arabic–English differences in token use and coding quality, the limits of using AI for advertising without marketing fundamentals, and how to choose a SaaS idea or learning specialization. Broader discussions address market and competitor research, unit economics, the difficulty of customer acquisition, and the value of starting with a focused market or specialty. They also include explicitly uncertain forecasts about autonomous agents, robotics, optional work, and large one-person companies, followed by personal—not universal—views on partnerships, accountability, disagreement, and the social effects of economic pressure.
87 moments from the stream
- 3:01:47
Competitor Research as Evidence of Market Demand
The speaker argues that competitors who have already reached customers can provide preliminary evidence that a market exists, while customer acquisition remains the central challenge. Before starting a project, he recommends researching competitors, their founders, websites, LinkedIn and Crunchbase presence, and investment signals, then looking for ways to offer something better or different. This is his view, not conclusive proof of demand.
- 3:06:32
A Structured Tool for Discovering Project Ideas
In response to a request about researching and choosing ideas, the speaker described a Markdown guide that used structured questions to help someone at the idea-discovery stage find a direction. He suggested that this approach could become a skill for generating ideas and identifying an initial direction, but presented it as a future possibility rather than a completed tool.
- 3:08:30
How to Start Learning and Building with AI
For people who feel lost because AI is evolving quickly, the speaker advises against assuming they have missed their chance. He recommends learning how to communicate with and use AI tools, then building small local projects for personal needs. He also argues that learning in a community can help people exchange knowledge and keep up with rapid changes.
- 3:10:51
Setting Search Campaign Budgets for Coverage, Not Just a Small Daily Cap
The discussion argues that a search budget should be evaluated as a strategic choice tied to bids and keyword coverage, rather than only as a fixed small daily amount. A search campaign may underspend its available cap, while a higher cap can cover a wider keyword set; this is general reasoning, not a reported final test result.
- 3:12:25
A Time-Qualified Comparison of Codex and Claude on Quality and Efficiency
Responding to a question about performance and token use, the speaker’s view at the time of recording was that Codex and Claude were close in quality, while Codex was more efficient and enabled more work. He stressed that this was a point-in-time assessment that could change as the tools evolve, and noted temporary Claude issues at that time.
- 3:13:37
Communicating an App’s Value Quickly on Its Website
A clear product-site goal is proposed: visitors should be able to understand what the app does within a few minutes. The emphasis is on fast comprehension and a clear value proposition.
- 3:14:45
Start a Marketing Agency with a Focused Specialty
The speaker advises someone considering a marketing agency not to launch every marketing service at once, because the field spans many specialties such as SEO, digital advertising, and social media. He recommends beginning with a clear specialty unless there is capacity to invest in a team covering multiple disciplines.
- 3:16:08
Keeping Project Files Independent of the AI Model
The speaker explains that a project need not belong to a particular AI model: its files can remain local or in a GitHub repository, allowing another model to access the same project, context, and skills when switching tools.
- 3:19:22
Waiting for Trials to Mature Before Judging Acquisition Results
The discussion explains that marketing results are delayed when users move from installation through a free trial and only then toward possible conversion. Evaluation should therefore wait for trial windows to mature and compare outcomes with conversion, retention, and churn assumptions, while recognizing that replacing churn requires ongoing acquisition. The stated rates were planning assumptions, not verified results.
- 3:21:04
Why Ad Results May Start Weak
The discussion says installs, trials, and feedback are important early on. It separately explains that initial Meta Ads or Google Ads results may be weak when there are too few data points, and that collecting more data can help the platforms identify likely interested users.
- 3:22:03
Researching Marketing Assumptions Before Starting
The speaker describes calculating expected metrics before starting the project and repeatedly revisiting them through further research. He says the research has reinforced the assumptions so far, while remaining eager to learn whether the metrics prove correct in actual use. Private figures and business details are omitted.
- 3:27:58
Using AI to Explore a Suitable Specialization
In response to a question about choosing a field to learn, the discussion suggests using AI repeatedly as a thinking partner and providing general context about skills and background to help explore a suitable direction.
- 3:29:34
Do Product-Description Keywords Fit Burnish’s Positioning?
A participant questions targeting terms such as “product description” and “product description generator” when Burnish is broader and offers that task among other capabilities. The exchange highlights the tension between a keyword that attracts real interest and search intent that may be narrower than the product’s full offering.
- 3:31:41
Choose the campaign objective before budget optimization
The speaker argues that choosing between ad-set- and campaign-level budget optimization comes after setting the correct core objective, criticizing a traffic objective when it does not match the intended outcome.
- 3:33:40
Relaying strategy across AI sessions and agents
The speaker demonstrates forking a session when work branches, then asking one session to relay an updated strategy to another session responsible for campaigns, reducing manual copy-and-paste. He also describes custom local agents used for task review, debugging, and code-coherence checks.
- 3:42:31
Executing a direct handoff between work sessions
The speaker shows the result of asking an updated plan to be sent to the session building the campaigns, presenting direct session coordination as an alternative to manually composing a handoff prompt and copying context.
- 3:43:37
A vision of goal-driven autonomous AI marketing
As a forward-looking opinion rather than a confirmed current capability, the speaker predicts AI moving from executing campaign instructions toward acting like a marketing executive: analyzing business history, creating assets and campaigns, monitoring performance, and pursuing goals set by an entrepreneur. The timeline and practical capability remain uncertain, and the transcript does not clearly establish the cited individual.
- 3:46:59
The prediction of large one-person AI companies
The speaker discusses a prediction that AI tools could enable one person to build an extremely valuable company, presenting it as a cited forecast rather than an outcome demonstrated in the excerpt.
- 3:48:15
APIs as the operating layer for enterprise agents
The speaker expects large-scale agent systems to rely on API consumption rather than fixed-limit subscriptions and predicts model-operation costs becoming a significant business expense; this is a forecast, not a tested result in the excerpt.
- 3:51:37
Should You Start a Business Alone or with a Partner?
A discussion of the trade-off between retaining full ownership and working with a partner whose skills complement the founder’s. Drawing on personal experience, the speaker argues that ending a joint venture need not end a friendship when both parties accept responsibility and avoid blame, while stressing that broad rules have exceptions.
Got a question about something in this stream?
Ask in the group