Description
Summary OVERVIEW We're looking for an experienced AI automation builder to design and build a system that generates high-converting landing pages by analyzing existing landing pages (structure, copy, and visuals) and recreating that same framework for our own brand – then publishing the finished page directly to Shopify. This is not a simple "AI writes copy" tool. It's a structural replication engine: given example pages, ads, and product research, the system needs to reverse-engineer *why* those pages convert (section architecture, copy framing, visual style) and rebuild that same architecture around our product and a specified marketing angle. WHAT AUTOMATION NEEDS TO DO 1. Input ingestion - Accept 1–5+ landing page URLs (or exported HTML) as structural/creative references – these may be from unrelated brands/industries, used purely for framework and framing, not niche relevance. - Accept supporting product/brand documentation (research docs, positioning notes, claims, mechanism explanations). - Accept existing ad copy (long-form) that is currently performing well, to be used as a secondary input for understanding proven angles and language. - Accept a specified marketing angle/awareness level to build the page around. 2. Structural + copy analysis - Break each reference landing page into its section-by-section architecture (hook, mechanism, proof, offer stack, etc.) – this needs to be detected programmatically, not just eyeballed. - Analyze the copy framing, tone, and psychological flow of each section (e.g., problem/agitation, unique mechanism, social proof, guarantee). - Cross-reference this against the provided ad copy and product research to determine what angle and claims are usable for our brand. 3. Visual analysis + recreation - Analyze the visual assets (images and video) used in each section of the reference pages – composition, style, format (UGC vs. studio, before/after, demo video, etc.). - Generate or source new visuals (image or video, matching the original asset type) that follow the same visual framework but feature our product/brand instead of the reference brand. - This is a hard requirement, not a nice-to-have – pages without matching visual replication are not acceptable output. 4. Output generation - Assemble the new page using the replicated structure, new copy, and new visuals. - Build the page in a Shopify-compatible format (we currently use GemPages) and publish/upload it directly to the Shopify store via API – no manual copy-paste step. IDEAL BACKGROUND - Proven experience building multi-step AI agent workflows (not just prompt chains) – e.g., n8n, Make, custom Python/LangChain/LangGraph pipelines, or similar orchestration frameworks. - Experience with LLM-based content analysis/extraction (structural parsing of web pages, section classification, HTML/DOM scraping). - Experience with AI image/video generation and/or editing pipelines (e.g., using generative models to produce on-brand marketing visuals from reference style inputs). - Working knowledge of Shopify APIs and ideally GemPages (or comparable page builders) for automated page publishing. - Direct-response / e-commerce marketing literacy is a strong plus – understanding of concepts like awareness levels, unique mechanisms, and long-form advertorial structure will make communication much faster. - Experience with web scraping/crawling tools for pulling reference page content and assets. DELIVERABLES - A working, documented automation (workflow files, scripts, or agent configuration – whatever platform you build it in) that I can run repeatedly with new inputs. - Clear input/output specification: what I need to provide each time, and what comes out the other end. - At least one full end-to-end test run producing a live, published landing page from a real set of reference inputs. - Basic documentation/loom walkthrough of how to operate and adjust the system after handoff. HOW TO APPLY Please include: 1. A short summary of a similar automation you've built (structural content analysis + generation + auto-publishing), with links/screenshots if available. 2. What platform/stack you'd propose for this (n8n, Make, custom code, etc.) and why. 3. Your approach to the visual recreation step specifically – this is the hardest part of the project and I want to know you've thought about it, not just the copy/text side. 4. Your availability and estimated timeline for a first working version.