Public AI product

MetaSculpt AI

A shipped AI product that turns crawl data, prompt context, citation tracking, and structured metadata generation into a usable workflow for search and AI-answer readiness.

Public Work Shipped 2024 - 2025
Visit project

Role

Co-Founder / Technical Lead

Tags

LLM IntegrationAI Product DeliveryStructured Outputs

Problem

People trying to compete in AI search needed a practical way to track AEO, GEO, and LLMO visibility, understand which prompts mentioned them, compare competitor citations, and inject useful Schema JSON into their sites.

Users

Site owners, marketers, and search/AI visibility teams trying to understand why they appeared or failed to appear in GPT, Perplexity, Gemini, and other AI-answer surfaces.

Ownership

  • Owned the AI workflow design, crawler logic, prompt tracking, experiment setup, weekly reporting, documentation, and code review with a co-founder.
  • Used a Wasp frontend template as a starting point, then drove the product workflow, data capture, report logic, and AI/search analysis behavior around it.
  • Made product and technical decisions jointly with the co-founder in a two-person build environment.

Hard Parts

  • Built a weekly report workflow that ran search checks across GPT, Perplexity, and Gemini, cleaned and stored results, extracted mentions and citations, and identified recurring competitors.
  • Crawled competitor sites that repeatedly surfaced in AI search, then compared those pages against relevant user pages to identify what competitors were doing better.
  • Turned prompt-level search results, citations, competitors, crawl findings, and site improvement opportunities into a readable recurring insight report.

Leadership

  • Led through execution rather than delegation: workflow design, documentation, code review, crawler/report logic, prompt tracker, and iterative product improvements.
  • Helped define the product model around measurable AI visibility rather than generic SEO advice.
  • Balanced startup speed with maintainability by separating crawler and generator responsibilities.

Shipped

  • AI search prompt tracker with competitor and citation capture.
  • Experiment tracking for monitoring how site changes affected prompt results.
  • Crawler-backed site and competitor analysis.
  • Weekly insight reports for visibility, competitors, citations, and recommended improvements.
  • Schema JSON generation and validation-oriented product flows.

Impact

  • Created a stronger process for turning AI search visibility checks into repeatable reports and product recommendations.
  • Improved the platform through direct involvement in workflow design, report quality, and competitor/site analysis loops.
  • The product shipped publicly, even though it did not ultimately gain user traction.

Highlights

  • Led product and technical architecture from prototype through public launch.
  • Built user flows for crawl status, schema generation, metadata recommendations, validation, and next steps.
  • Separated crawler and generator responsibilities to reduce coupling and improve maintainability.
  • Designed the Crawl -> Generate -> Validate -> Fix workflow model so non-specialists could act on complex AI/search concepts.
  • Closed the gap between idea, product behavior, interface, documentation, and deployment planning without a dedicated engineering team.

Tools

OpenAI APIGemini APISchema.orgREST APIsPrompt contextStructured metadataValidation workflowsReact/Next.js conceptsTypeScriptDockerAWS-oriented deploymentRAG/context-cache concepts