Skip to content
← Back to Portfolio
Precek — AI for text, image, audio and video
Precek — AI for text, image, audio and video

Precek — AI for text, image, audio and video

A web app that processes text, image, audio, and video with AI models: visualizations, summaries, trends, and data export.

Project date: April 10, 2024

Live DemoView Code

AI Highlight

  • Context:A web app that processes text, image, audio, and video with AI models: visualizations, summaries, trends, and data export.
  • Technologies:Next.js, React, TailwindCSS, Typescript, MUI, IndexedDB
  • Summary:# Context A lot of content (PDF, EPUB, image, audio, video) ends up as “a file on disk”: hard to compare, combine, and spot patterns. Precek treats media as data — it processes files with AI models and builds a temporary...

Precek — dashboard

Context

A lot of content (PDF, EPUB, image, audio, video) ends up as “a file on disk”: hard to compare, combine, and spot patterns. Precek treats media as data — it processes files with AI models and builds a temporary database for visualizations, summaries, and export.

Goal

I designed and delivered a tool that:

  • processes text, image, audio, and video via AI models (OpenRouter / OpenAI),
  • builds a temporary database for analysis and visualization,
  • generates summaries, trends, contexts, and word clouds,
  • exports results to CSV.

My role

I owned design, solution design, implementation, and up to 30 days of post-launch care.

Functional scope

  • AI media processing — images, audio, video, and text,
  • Database management — temporary store for visualization,
  • CSV export — download data for external analysis,
  • Data visualization — summaries, trends, contexts, word cloud,
  • Cross-platform — web app with a mobile direction (React Native in the project stack).

Approach

I deliberately chose client-side processing + IndexedDB: working data does not have to land in the app’s cloud by default. AI models connect through OpenRouter (including lower-cost / free paths) or OpenAI depending on configuration. The frontend (Next.js + React + TypeScript + Tailwind/MUI) links upload flows to visualizations (D3, force graph, word cloud) and document parsers (PDF.js, epub.js).

Challenges

  • Many media types, one pipeline — each format enters differently; results had to normalize into a shared data model for visualization.
  • Model cost and access — OpenRouter as a model-choice layer without hard lock-in to a single provider.
  • Working-data privacy — IndexedDB as a session/temporary store instead of “everything on the server” by default.

Outcome

The result is a live app (demo, GitHub) that turns media into an analyzable dataset with visualizations and CSV export. I provided up to 30 days of post-launch care after go-live.

Features

Mobile view

Analytics

Stack

  • App: Next.js, React, TypeScript, React Native (mobile direction)
  • UI: Tailwind CSS, MUI, NativeWind
  • Visualization: D3, React Force Graph, React WordCloud
  • Documents: pdfjs-dist, epub.js
  • Data: IndexedDB
  • AI: OpenRouter API / OpenAI API

Related projects