A 70 isn't an ugly email. It's usually a fine-looking email with one structural flaw that compounds — a CTA that blends into the background at small sizes, a two-column layout that stacks badly on Android, or a content hierarchy where three things compete for attention at the top and the reader picks none of them.
I built kopi to catch exactly that class of problem before send. Every generated email gets an automated critique across 12+ criteria: layout hierarchy, mobile readability, CTA contrast, whitespace distribution, and others — each scored 0–100, aggregated into a single number. The score isn't decorative; it flags the specific criterion and explains why it lost points.
Generation takes a text prompt — campaign goal, brand tone, audience segment — and returns a complete HTML email in under 5 minutes. The critique runs on every output automatically. You don't request a review; it's part of the loop.
The trykopi.ai/emails gallery has 400+ emails that scored 80 or above, which is the fastest way to see what the rubric rewards in practice. A 93 typically has one clear visual path, a CTA that passes contrast on a phone screen in bad lighting, and a mobile stack with no orphaned lines or collapsed images.
The tool integrates directly with Klaviyo for export and works as an MCP server inside Claude Code, Cursor, and ChatGPT — so if you're already living in an editor, the generate-critique-export loop doesn't require a context switch.
Top comments (0)