UI RENDERING & SSR FEATURE
Submit AI Task
Enter text to process through the 4-stage AI pipeline.
Feature Guide
A quick orientation block that answers the essential questions: what this feature does, how it works, why it matters, and the key concepts behind it.
What this does
Submit a task and watch the AI pipeline stages reveal one by one as the cron job processes it.
AI Pipeline Input
Submit text for processing
Enter text below and submit it into the four-stage AI pipeline.
Verified against Semitexa Ultimate 2026.09.19.1020
Reactive AI Task
Submit a task and watch the AI pipeline stages reveal one by one as the cron job processes it. The page stays SSR-first while background AI stages keep advancing on the server, so the sidebar and the live pipeline share one consistent page shell.
How it works
The user submits text through a form. The handler creates a DemoAiTask record with status pending. A cron job picks up pending tasks and processes them stage by stage, writing stage results into the task record as JSON. A reactive slot with refreshInterval: 2 holds its SSE connection open while the server re-renders the pipeline view every two seconds and pushes it, reading the latest stage results from the task record and showing each completed stage.
The flow
- User submits the form — a task record is created with
status: pending. - The cron worker picks up the task and starts advancing through stages.
- Each stage completion is written to
stage_resultsas JSON. - The reactive slot refreshes every two seconds, reading the updated stage data and rendering the current pipeline state as HTML.
- When all stages are complete, the slot shows the finished result.
Key mechanisms
DemoAiTask— the task record that carries status, stages, and stage results.- stage-by-stage — the cron processor advances one stage at a time and writes intermediate results.
refreshInterval: 2— the server re-renders and pushes every two seconds; the browser never polls.- user-triggered → cron pickup — the form submission creates a task; the cron picks it up without a direct async handoff.
Why this matters
AI pipeline progress is a case where client-side state simulation is especially misleading — AI stage durations are unpredictable, so any client-side estimation is just fabricated progress. Letting the server own the stage results and reflecting them through a live slot gives the user accurate information throughout the processing run.