How to use Cursor to build a custom SEO prototype
Hans Brattberg used Cursor Agent mode to build an SEO prototype with a database and GitHub deployment, approving it and refining its interface from screenshots.
abundly.ai
Original prompt
Shared by the source author. See source notes for context.
I want this deployed on Vercel. With a database. Automatic deploy via GitHub. What's the easiest way?
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- Hans Brattberg
- Original account
- Read the source ↗
- Published
- Source link checked
- Reported result
- Working prototype demonstrated to colleague.
- Human involvement
- Author approved proposed stack and iterated interface through screenshots.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- Asked agent for Vercel/database/GitHub deploy setup.
- Agent proposed and installed Next.js/PostgreSQL/Docker/pipeline.
- Author supplied screenshots and UI changes.
Hans Brattberg wanted a simpler way to inspect SEO information after finding existing tools difficult to use. He opened Cursor in Agent mode and asked for a deployment on Vercel with a database and automatic deployment through GitHub.
The agent proposed Next.js, PostgreSQL, Docker and a GitHub Actions pipeline. Brattberg approved the choices and describes Cursor installing the setup and working through implementation. He also used natural-language requests for Git operations rather than issuing every command himself.
The following morning, he iterated on the interface by supplying a screenshot and describing the desired change. Cursor made those revisions in the project context. Brattberg then showed a screenshot of the result to a colleague, who mistook the prototype for an existing commercial product.
The outcome reported here is a working SEO-tool prototype, not an improvement in rankings or a production SaaS launch. Its transferable pattern is concrete deployment requirements followed by visual feedback, with the human accepting the stack and judging the interface. The source does not supply a performance or SEO-effectiveness evaluation.
Prompt
Original prompt
Source: Blog · Hans Brattberg
Frequently asked questions
What did the initial request specify?
Brattberg requested Vercel deployment, a database and automatic deployment through GitHub, then asked for the easiest setup.
How did he refine the interface?
He supplied screenshots of the current interface and described how he wanted it changed; Cursor implemented the revisions.
What result does the account establish?
The author reports a working prototype demonstrated to a colleague. It does not establish improved search rankings or a production service launch.
More details
Remaining steps
- Agent managed push/deploy steps.
About this account
Cursor Agent mode proposed and installed a stack, implemented the prototype and revised the interface from screenshots and human direction.
Primary first-person account opened and checked on 2026-10-14. Prototype report; no measured search-ranking improvement. Outcomes and test claims are attributed to the source and were not independently reproduced. Prompt is a short verbatim source request.
Credit: Hans Brattberg on Blog
Source published
Added to Agents In Use
Source checked
Tools and topics
Cursor · Next.js · PostgreSQL · Docker · GitHub Actions · Vercel
SEO prototype · Agent mode · Screenshot feedback
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.