Cursivis

Cursor-native intelligence for Logitech MX hardware

Selection becomes context. Logitech hardware becomes intent. Cursivis turns both into action.

Cursivis reimagines the cursor as an intelligent workflow surface. Instead of opening a chatbot, typing prompts, and switching between apps, the user simply selects what matters and triggers Cursivis through MX Master 4, MX Creative Console, or the Actions Ring so the system can respond directly inside the flow of work.

Selection-first interaction Context-aware reasoning Live browser execution Haptics + hardware feedback

The system understands the current context, returns the intended result, and can carry that result forward into the live browser workflow. The interaction stays simple. The intelligence adapts to context.

The same deliberate trigger can summarize a long article, translate foreign text, explain code, debug broken logic, read visual content, or continue a browser task through Take Action.

Cursivis concept art showing cursor-native intelligence

CORE INTERACTION MODEL

Selection + Trigger + Context = Action

Cursivis begins with what the user is already working on. The selection carries the context. Logitech hardware expresses deliberate intent. Cursivis bridges those two moments and responds with the most relevant next step.

Select Trigger Understand Generate Execute

PITCH + DEMO

See the system in motion

The idea video explains the interaction shift. The execution video shows the full workflow in action.

Idea video

Watch on YouTube

The cursor needed an upgrade

This pitch frames the problem: intelligence is powerful, but the current interaction model still creates friction.

Execution demo

Watch on YouTube

Selection to real browser execution

This demo shows Smart Mode, Guided Mode, Snip-it, text refinement, and Take Action in a live browser workflow.

AUDIO PITCH

Listen to the narration

INTERACTION SHIFT

What Cursivis changes

AI usually begins with a blank box. Cursivis begins with the thing the user is already working on.

The current default AI interaction asks the user to describe everything first. Cursivis starts with the thing the user already selected.

The user should not have to stop working just to explain context to a model. In Cursivis, the selection already carries that context. Logitech hardware provides the deliberate trigger. Cursivis bridges those two moments and returns the most relevant next step.

The same physical action can therefore produce entirely different outcomes depending on what is selected: a long article becomes a summary, foreign text becomes a translation, broken code becomes a debugging flow, and a live browser task can turn into real follow-through through Take Action.

Illustration of prompt friction being replaced by direct cursor-native interaction — From prompt friction to direct, hardware-native intent.

CORE MODES

Smart Mode

Automatically chooses the best action for the current selection.

Guided Mode

Lets the user navigate context-aware options with Logitech controls.

Talk / Text Trigger

Adds refinement through voice or a floating prompt bar.

Snip-it

Handles image regions, OCR, object understanding, and color capture.

Take Action

Converts results into structured browser execution steps.

Context-aware Cursivis workflow branching from the cursor into multiple actions

PROOF OF CONTEXT

How the system responds to selection

The interaction stays constant. The output changes with the context already on screen.

Cursivis translating selected French text

Summarize and translate

Long passages, rules, and foreign-language text become useful output without leaving the current workflow.

Explain and debug code

The same trigger can explain logic, diagnose problems, and move toward repair depending on the selected code.

Cursivis understanding a selected image region

Understand visual content

Snip-it extends the workflow into OCR, image description, object recognition, and color capture.

Cursivis executing a browser task with Take Action

Carry results into action

Take Action turns generated output into live browser follow-through for forms, MCQs, and repetitive tasks.

WORKFLOW

Workflow at a glance

The front-stage experience stays compact: select, trigger, understand, return the result, and optionally act.

LIVE INTERACTION PATH

From selection to follow-through

Workflow diagram

flowchart LR
    A["Select live context<br/>text, code, image, form, email"] --> B["Trigger from Logitech hardware"]
    B --> B1["MX Master button"]
    B --> B2["Actions Ring"]
    B --> B3["MX Creative Console"]
    B1 --> C["Cursivis captures context"]
    B2 --> C
    B3 --> C
    C --> D["Smart Mode or Guided Mode"]
    D --> E["Result appears in the AI box"]
    E --> F["Insert / More Options / Talk"]
    E --> G["Take Action"]
    G --> H["Execute in the live browser workflow"]

Select

Choose live context: text, code, image, form, or email.

Trigger

Use MX Master, Actions Ring, or MX Creative Console.

Understand

Cursivis captures context and routes the task.

Return Result

Smart Mode or Guided Mode returns the right response.

Act

Insert, refine, or execute the next browser step through Take Action.

TECHNICAL STACK

System architecture

A local-first stack where hardware input, orchestration, reasoning, execution, and feedback stay tightly aligned.

STACK OVERVIEW

How the system is layered end to end

Architecture diagram

flowchart TB
    A["Logitech Hardware<br/>MX Master 4 / MX Creative Console / Actions Ring"] --> B["Logitech Plugin Layer<br/>C# + Logi Actions SDK"]
    B --> C["Shared IPC Protocol"]
    C --> D["Cursivis Companion<br/>C# + WPF"]
    D --> D1["Selection capture"]
    D --> D2["Orb + AI box UI"]
    D --> D3["Smart / Guided / Talk / Snip-it / Take Action"]
    D --> E["Reasoning Backend<br/>Node.js"]
    E --> E1["Context understanding"]
    E --> E2["Action routing"]
    E --> E3["Structured browser plans"]
    D --> F["Browser Action Layer"]
    F --> F1["Chromium extension"]
    F --> F2["Browser action agent"]
    F --> F3["Native host / fallback execution"]
    E --> F
    F --> G["Real browser tab and live workflow"]
    B --> H["Haptics + hardware feedback"]
    D --> H

Selection capture Orb + AI box Multimodal routing Structured browser plans Live follow-through

HARDWARE ADVANTAGE

Why Logitech MX changes the interaction

Logitech does not just launch the workflow. It gives Cursivis the physical language of intent.

Precise pointing

The cursor becomes the capture point for text, visuals, UI, and live browser context.

Trigger + thumb wheel

Deliberate buttons and the horizontal wheel make Guided Mode feel hardware-native and fast.

Actions Ring layering

Nested actions provide fast reach into Trigger, Talk, Snip-it, Take Action, and Settings.

Haptics

Tactile feedback confirms action change, execution, and completion without stealing visual attention.