$49 Reservation Funnel
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 documents Burnish’s preparation before market entry, including practical reviews of the website messaging, mobile experience, and the “scan my store” journey intended to measure a Shopify store’s visibility in AI-platform answers and email a useful report. Work focused on outcome-led copy, reframing the product loop as “research, monitor, action, measure,” and experimenting with sections, ambient video, and visual effects whose impact on site speed still needs measurement. Refundable early reservations and a waitlist were discussed as possible demand-validation signals, while fabricated social proof was rejected and actual uptake remained uncertain. Technical discussion covered the Polaris migration, dividing work between Codex and Claude, deployment and environment-configuration complexity, and adding model providers; however, the migration audit, compliance requirements, and some sensitive backend integrations remained unresolved. Side discussions answered audience questions about learning English, stream-music permission, the current preference for YouTube, and a multi-agent command-center architecture. The recording is presented as documentation of an unfinished build journey—not a course or a claim of product success.
50 moments from the stream
- 1:12:18
Reviewing the ScanMyStore CTA Flow and Section Structure
Open decisions were to be documented before reviewing the design, end-to-end process, and funnel for the ScanMyStore call to action—what existed and what remained. An initial look at the new section structure produced with Claude was judged clearly improved, but the captions do not establish that the flow was complete.
- 1:13:43
Designing a Scroll-Reactive Video Background
The grid background was reconsidered in favor of a subtle, minimal video that advances frame by frame with scrolling, matches a stone-paper palette and texture, and visually supports the section’s core concept.
- 1:17:11
Keeping Sensitive Integrations Off-Stream
The discussion explains that some reservation-funnel work, such as connecting API keys and a payment service, is better completed off-stream to protect sensitive information, while safe portions of the site and funnel can still be developed publicly.
- 1:21:19
Getting Permission and Crediting Stream Music
In response to a question about the stream music, he said he contacted its creator for permission because it is copyrighted and credits the creator’s channel in the video description.
- 1:24:30
Experimenting with an AI-Generated Ambient Video Loop
After dissatisfaction with the results, the strategy shifted to generating an AI image or video and animating it as a seamless loop with matching start and end frames. The goal was a quiet ambient background that would not obviously read as video, while acknowledging that the final visual direction was still undecided.
- 1:26:46
Ambient Background and Separate Image/Animation Prompts
The section’s stark white space prompted an idea for a subtle ambient background, followed by a request for separate image-generation and animation prompts describing the desired look and motion without mentioning duration or start and end frames.
- 1:29:55
Balancing Visual Effects with Site Speed
The discussion proposes measuring site speed after adding visual effects rather than assuming their impact, noting that whether an MP4 is acceptable depends on the page’s overall weight and that the current build is heavier than an earlier comparison site.
- 1:33:35
Choosing and Optimizing the Media Format
MOV was considered too heavy, so the provisional choice was an optimized MP4, with a later check for conversion to a lighter web-oriented format; the exact final format name is unclear in the captions.
- 1:35:07
Turning Deferred Work into Milestone Tasks
Work that would not be handled immediately was to be broken into tasks and placed in the relevant milestone, preserving follow-up while attention shifted to another priority.
- 1:39:51
Designing the Store-Scan Journey from Click to Email
The work focuses on both the appearance and full flow of the primary “scan my store” call to action—from the click, through the form page, to receipt of the email—rather than limiting design attention to the report itself.
- 1:41:50
Comparing an API Reseller with Direct Access
The speaker said he discovered the service was a third-party intermediary reselling API access and believed direct use through Google AI Studio could provide the same service more cheaply. This was his live assessment, not a documented comparative test.
- 1:42:34
A Premium Vertical Ambient-Motion Direction
A visual direction was defined for a purely ambient, premium-looking video with vertical motion, explored through a small set of concepts with image and animation prompts for each.
- 1:42:34
Brainstorming When the Visual Direction Is Unclear
Because the desired look was still unclear, the brainstorming was constrained to three ideas at a time, each with image and animation prompts plus a preferred option, before narrowing the work to polishing the scan-form page first.
- 1:44:15
Using a Store Audit Report as a Reservation Funnel
A proposed section collects a Shopify store URL and business email, performs a real scan, and emails a report about the store’s visibility on AI platforms. The site could show a teaser, while a useful report may encourage the merchant to return, subscribe, or place a reservation; the speaker notes that this requires substantial backend and API work.
- 1:47:15
Designing Scroll-Reactive Video Motion
A visual asset is planned to react frame by frame to scrolling, so it does not need to loop. The proposed workflow moves directly to a video prompt, with particles animated vertically and concentrated toward the sides rather than the center.
- 1:51:56
Architecture of a Multi-Agent Command Center
The command center includes an Agent Hub for viewing agents, their models, and sub-agents. Commands launch suitable groups of agents, which report into shared JSON state so each model can access the relevant portion. Similar ideas may be found on GitHub under terms such as “command center” or “mission control,” though implementations are typically customized to their owner’s workflow.
- 1:53:07
Adding Custom Agents to the Command Center
New agents can be connected through MCP, designed for a specific role, added to the Agent Hub, and activated by a command. This is described as an available capability that the speaker had not yet enabled.
- 1:56:41
Presenting AI Engines with a Shuffling Icon Effect
A UI section is planned to present six AI-engine logos, reuse an existing icon-shuffle effect, and prevent duplicate logos. The speaker prefers visually filled sections without awkward side gaps, then selects one proposal and asks for it to be implemented.
- 1:57:37
Deferring Hermes for Future Automation
The speaker sees potential future value in Hermes for customer-service automation, debugging, and possibly development work, but defers it while focusing on the current build phase.
- 2:05:41
Redesigning the Homepage Audience and Cohort Sections
A review of the “Who it’s for” and cohort sections near pricing, including a proposal to combine them into an experience that helps visitors find their fit and to explore several design directions before implementation. The mobile preview also showed that the page was not yet optimized and that the current section needed a broader redesign.
Got a question about something in this stream?
Ask in the group