Skip to content
Sercan — Hüseyin Sercan Gündoğan

Case study

Seoneer

An autonomous SEO engineer for GitHub-hosted Next.js projects — it understands the product and codebase, picks one high-value safe action, and ships it as a pull request.

End-to-end ownership
Seoneer product overview — autonomous SEO engineer for GitHub repos
Placeholder asset — Product overview

Overview

Seoneer is an independent SaaS product that acts as an autonomous SEO engineer for software projects. It connects to a GitHub repository, builds product intelligence from the codebase, and executes the highest-value safe SEO action as a reviewable pull request — not a content farm.

Context

I built Seoneer as a solo product to solve recurring SEO work for technical founders who already ship on GitHub and Next.js, but lack time for sustainable organic growth.

My role

I own product direction, architecture, implementation, agent contracts, billing, and iteration as creator and full-stack product developer.

Ownership

End-to-end ownership of the modular monolith: auth, GitHub App integration, repo analysis, SEO agents, Trigger.dev jobs, Search Console, payments, notifications, and the operations UI.

Problem

Traditional SEO tools produce dashboards and recommendations. Generic AI content tools produce low-trust articles. Founders need shippable, reversible repository changes grounded in product context — with human review by default.

Contribution

  • Designed a people-first action catalog (titles, links, technical SEO, articles) with explicit NO_ACTION / WAIT stop conditions.
  • Built repository analysis and versioned Project Intelligence Profiles from Next.js App Router codebases.
  • Orchestrated multi-stage agents (analyst, strategist, researcher, writer, code agent, reviewer) via Trigger.dev.
  • Delivered every normal change as a GitHub PR with quality gates — never writing to the default branch by default.
  • Integrated GSC signals, Dodo credit billing, Better Auth, and Resend approval notifications.

Technical decisions

  • Modular monolith on Next.js App Router rather than premature microservices.
  • Domain logic in feature modules; long AI/crawl/build work only in Trigger.dev tasks.
  • Structured Zod-validated agent I/O over a provider-agnostic Vercel AI SDK facade.
  • Drizzle + PostgreSQL as source of truth; GitHub App for safe repo mutations.

Challenges

  • Keeping autonomous SEO safe: path policies, human review, and explainable decisions.
  • Selecting one highest-value action per cycle without incentivizing content spam.
  • Running expensive analysis outside HTTP request lifetimes.
  • Grounding AI output in real product evidence instead of fabricated claims.

Outcome

  • A shipping independent SEO operations product at seoneer.site with PR-first delivery.
  • Demonstrates end-to-end ownership of AI agents, GitHub automation, billing, and product UX.

Visuals

Seoneer operations dashboard with selected SEO action and pipeline
Placeholder asset — Operations dashboard

Technologies

  • Next.js
  • TypeScript
  • React
  • Drizzle ORM
  • PostgreSQL
  • Better Auth
  • Trigger.dev
  • Vercel AI SDK
  • GitHub App
  • Google Search Console
  • Dodo Payments
  • Resend
  • Zod
  • Tailwind CSS

Links