BACK TO CASE FILES
CASE FILE 002 // ARCHIVE LOG

DivasAI

Generative AI Platform for Designers

01 / The Challenge & Problem

Designers spend hours writing prompts and waiting for image generations without contextual file version control. DivasAI needed to create a collaborative interface to test, cache, and refine AI image outputs in real-time.

02 / Discovery & Research

We discovered that prompt engineering workflows are highly iterative. Designers needed to view history trees like Git. Additionally, API fees were high due to repeated image generation of identical prompt contexts, making semantic caching essential.

03 / Construction & Development

We constructed the user interface with Next.js and Framer Motion, designing an interactive visual canvas that tracks prompt revisions. We implemented a vector caching algorithm that reduced external OpenAI API expenses by 42%.

SYSTEM COMPONENTS

Next.jsExpressPineconeOpenAI APITailwind CSSMongoDB

RESULTS RECORDED

42% savings on API generation expenses, 80% reduction in average wait times for cached images, and loaded with a 98/100 Google Lighthouse speed index.

SCHEMATIC DRAWINGS

Visual Blueprint & Metrics

LOG # DIVASAI
D

DivasAI Case File

SYSTEM SCHEMATIC SUMMARY

Next.jsExpressPineconeOpenAI APITailwind CSSMongoDB
Chat on WhatsApp