What's New

The latest updates and milestones as we build Procera.

Frontend Application: Complete Instruction Management Suite

We've launched the full Procera frontend β€” a clean, purpose-built web application for managing workflow instructions end-to-end.

  • Instruction management β€” browse, organize, and navigate all generated instructions in one place.
  • Refinements β€” submit improvement requests either as a general comment across the full instruction or as a targeted note on a specific step.
  • Inline editing β€” adjust instruction content directly in the interface without leaving the page.
  • Export β€” download any instruction as a formatted PDF or DOCX file, ready to share with your team.

Backend Preprocessing Pipeline

We introduced an intelligent preprocessing layer that sits between the desktop agent and the AI model. Raw captured data is automatically cleaned, deduplicated, and restructured so the model receives properly formatted, contextually rich input every time.

This step was critical for improving the consistency and accuracy of generated instructions β€” especially for complex, multi-application workflows where raw event streams can be noisy or redundant.

Desktop Recording Agent

The Procera desktop agent is now live. It's a lightweight background process that automatically captures everything needed to generate a high-quality instruction: UI element state, user actions, application context, click coordinates, and timing metadata.

The agent works silently without interrupting the workflow being recorded β€” no manual annotation, no stopping and starting. Just perform the process once, and Procera handles the rest.

Core Platform: Instruction Generation & Management

We shipped the foundation of Procera β€” the core platform powering the full instruction lifecycle from capture to delivery.

  • Instruction generation β€” AI transforms recorded workflow data into structured, step-by-step SOPs.
  • Refinement system β€” instructions can be iteratively improved based on user feedback.
  • Instruction library β€” all generated instructions are stored, versioned, and retrievable.
  • Visual context β€” each step includes a full-page screenshot, a cropped view of the relevant area, and a click-position overlay marking exactly where the action occurred on screen.