Escribolt

Ask Threads turn your static database of transcripts and notes into an interactive memory graph. Instead of manually searching, reading, and summarizing long-session audio recordings or extensive project directories, you can launch a conversational chat thread powered by local or BYOK LLM models.

Context Selection & Inline @ Mentions

An Ask thread can be general or targeted. You can append specific context to your questions in two ways:

  • Context Selection Dropdown: At the bottom of the chat composer, use the context selector button to lock the conversation to a specific Note, Recording, Folder, or Chat. The composer placeholder will update to indicate your selected scope (e.g., "Ask about note 'Design Feedback'...").
  • Inline @ Mentions (Dynamic Context): While typing your prompt, type the @ symbol to trigger a recommendation popover. This list displays your database items matching your query. Selecting an option inserts a structured reference token, such as:
    • @note:"Meeting Briefing"
    • @recording:"Marketing Sync 2026-06-04"
    • @folder:"Product Specs"
    • @chat:"Tech Architecture Q&A"
    Escribolt automatically compiles the full text of all mentioned documents, formats them as structured background data, and appends them to your query prompt.

Reasoning & Thinking Traces

When you submit a query, the AI model generates a step-by-step Thought Trace before writing its final response.

  • Click the collapsible Thought header above any streaming AI response to view the step-by-step execution path.
  • The trace displays real-time milestones (e.g., "Reading recording transcript...", "Summarizing context...", or "Preparing templates...") to show you exactly how the LLM is gathering and analyzing the selected context items.

Interactive Citations & Tooltips

AI responses remain grounded in your source materials by embedding interactive, numeric citations (e.g., 1) directly into the generated text.

  • Hover Tooltips: Hover your mouse cursor over any citation tag. Escribolt displays a popup showing the title and snippet metadata of the referenced source.
  • Direct Navigation: Click any citation tag to instantly open that note, folder, or recording in a new workspace tab on your dashboard.

AI Thread Capabilities & Actions

Ask Threads support several actions to turn AI answers into permanent knowledge:

  • Save as Note: Click the Save as Note action on any assistant message block. Escribolt automatically extracts the response text and creates a new, editable rich text note in your sidebar directory.
  • Formatted Clipboard Copy: Copy chat answers with full styling. Escribolt exports both a clean Markdown text copy and a rich HTML formatted blob, allowing you to paste styled text directly into applications like Apple Notes, Microsoft Word, or email clients.
  • Automatic Title Generation: When starting a new session, Escribolt automatically prompts the active language model to generate a descriptive title based on your first message. You can also manually rename or delete threads from the sidebar.

Hands-Free Voice Inputs

Escribolt's Ask tab is designed for quick, keyboard-driven inputs, but you can also use your voice to ask questions:

  • Focus your cursor inside the chat composer.
  • Use your configured macOS system-wide Live Dictation shortcut (e.g., holding the Fn key or pressing Fn + Space).
  • Speak your question. Escribolt transcribes your query in real time directly at the cursor and inserts the text. Press Enter to submit.