Global summon shortcut: press Ctrl+Shift+Space from any Windows application to bring the assistant forward instantly, with no window to search for and no tab to reload.
Customizable key combination: change the shortcut in Settings under Keyboard, using modifiers with Space, a letter, or a number, such as Alt+Space.
Conflict handling: if Windows cannot register a chosen combination because another application already holds it, SideNote Pro notifies you and restores your previous working shortcut.
Right-edge docking and drawer leash: anchor the panel to the edge of your current display and retract it into a small 44 by 68 pixel leash when you step away from it.
Position memory: clicking the leash restores the exact size and vertical position you left the panel at, so nothing needs re-arranging.
Display change resilience: the leash re-docks itself automatically when Windows reports a resolution change, a work-area change, a DPI change, or a display disconnect.
Flexible window modes: use SideNote Pro as an ordinary floating window, maximize it, or keep it docked, depending on what suits the task in front of you.
Wide layout sidebar: widen the window past 1,200 logical pixels to keep your conversation history visible in a sidebar while answers keep a comfortable reading width.
Independent always-on-top: the assistant window can stay above other windows on its own, and the drawer leash can stay on top independently without forcing the full window to be topmost.
Bring Your Own Key AI connections: configure OpenAI with your own secret key, or connect a local Ollama or LM Studio endpoint so inference runs entirely on your own hardware.
Any compatible endpoint: set up another OpenAI-compatible service manually, keeping in mind that compatibility varies between services and is worth checking beforehand.
Per-profile model selection: choose the specific model for each provider profile, plus an optional reasoning effort level where your chosen model supports one.
Document input: attach PDF, DOCX, TXT, and Markdown files directly into a conversation and ask questions about their content.
Local text extraction: text is pulled out of the document on your own machine, and the original file itself is never uploaded anywhere.
Screenshots and image capture: capture a region, a specific window, a full screen, or the entire virtual desktop, or paste directly from the clipboard.
Drag and drop support: drag an image file straight into the conversation window instead of capturing a new screenshot each time.
Local folder context: point SideNote Pro at a folder you choose and ask questions about the files inside it.
Lexical search, not crawling: nothing is discovered or watched in the background, and indexing happens locally with no AI provider contacted and no embeddings or vector database involved.
Attached sources: every answer keeps a list naming the files and PDF pages it drew on, so you can open and verify the original material yourself.
Selected-text actions: summarize, rewrite, explain, or translate a text selection in any Windows application using a shortcut.
Custom actions: define your own selected-text actions beyond the built-in set to match tasks you repeat often.
Inline commands: type a short trigger such as /fix, /en, /short, or /formal followed by your text in any Windows application, then press Tab to have the AI’s version typed back in place.
Custom commands and writing context: add your own commands and set a writing context that applies to all of them, with every command switched off until you turn it on.
Local conversation history: threads, search, pins, and retention settings are stored entirely on your device, with no cloud synchronization involved.
Markdown and PDF export: export an entire thread as Markdown, or export a single assistant response as a formatted A4 PDF rendered locally.
Usage and cost tracking: see token counts and local cost estimates for the AI usage you generate.
Daily ledger and limit: track a running daily ledger of usage and optionally set a daily limit to keep costs predictable.
Response rendering and controls: answers render as Markdown with copyable code blocks, and you can stop, continue, regenerate, or edit a response as needed.
