SCND.AI User Guide

SCND.AI is a private work-memory app for people who want their daily work to become searchable context for themselves and for AI agents. It watches your workday, turns activity into structured episodes, extracts important entities, and can expose that context through a local MCP server.

This guide covers SCND.AI v0.8.0 as observed in the desktop app.

Core Idea

SCND.AI does not work like a regular notes app. You keep memory running while you work, and the app builds a structured record of the day:

what happened during each work episode
which apps and topics were involved
which people, files, projects, websites, tools, or concepts appeared
what open questions remain
how focused the day was
what context can be shared with AI agents through MCP

The most useful results usually appear after SCND.AI has collected enough activity. The onboarding suggests that first insights normally need about 30-60 minutes of work. Daily images and richer summaries may require several hours of activity.

First Setup

Open SCND.AI and check the memory status in the sidebar.
Use the red/green memory control to start or pause capture.
Open Settings -> Privacy and choose how memory starts:

- Auto: starts automatically. - Manual: you start it yourself. - Remind me: SCND.AI reminds you if memory is still paused after launch.

Configure the privacy filter before using SCND.AI for real work.
Review storage limit and memory resolution.
Fill in User profile if you want more personalized summaries and agent context.

Navigation Overview

Area	What It Does
Dashboard	Daily overview with focus, timeline, apps, goals, entities, and MCP status.
Journal	Searchable timeline of work episodes, actions, entities, and open questions.
Memory graph	Private wiki of entities seen across your work.
MCP	Local Model Context Protocol server for AI agents.
Analytics	Productivity metrics across today, yesterday, week, month, or a picked date.
Goals	Daily goals with AI-assisted verification against real work.
Assistant	Hotkeys and rules for Notice and Surface awareness modes.
Settings	Privacy, capture, model, productivity, profile, subscription, and appearance.
Burger menu	Dark mode, product tour, docs, and exit.

Dashboard

The Dashboard is the best daily starting point. It shows:

current date and whether you are viewing today or a past day
tracked time and deep-work time
focus score
productive and distraction time
daily archetype and narrative summary
key bullets for the day
lingering insight and open question
top apps and top entities
activity timeline
focus session
daily goals
image of the day, when enough activity is available
MCP status

Use the calendar to review previous days. Colored days indicate available data and approximate focus score. Past days show completed summaries, timelines, and top apps/entities when analysis is available.

Journal

Journal turns your day into episodes rather than raw screenshots. Each episode may include:

title and time range
short gist
what you did
entities and artifacts
open questions
linked sources for later review

Use the date tree to move between days. Use search to find words across episodes, entities, actions, and questions. The search shortcut is shown as Cmd+K.

The right-side structure panel helps scan a day by:

section outline
open questions
entities mentioned that day

Open questions can be marked as resolved or unresolved. This makes Journal useful as a lightweight follow-up list, not just an archive.

Journal also includes an Agent tab for asking questions against your memory. Because answers may use your captured work context, review your model and privacy settings before relying on it for sensitive information.

Memory Graph

Memory Graph is SCND.AI's private wiki. It groups extracted entities into types such as:

tools
projects
concepts
libraries
files
companies
events
functions
people
websites and organizations, where available

You can search, filter by type, and sort by recent or popular entities. Opening an entity shows:

description
aliases
mention count
key facts
open questions
contradictions, if found
related entities
sources linked back to Journal episodes

Use Memory Graph when you need to remember where a project, person, file, or tool appeared and what SCND.AI knows about it.

MCP

The MCP screen exposes SCND.AI memory to external AI agents through a local server.

When enabled, the app provides a localhost endpoint like:

http://127.0.0.1:8787/mcp

The screen provides:

MCP on/off toggle
endpoint copy button
bearer token copy button
Codex config copy button
token regeneration
port setting, editable only when MCP is stopped
available tool descriptions

Observed MCP tools include:

search_context_memory
get_day_context
get_episode
get_recent_context
find_entity_mentions
list_open_questions
get_entity_profile
list_top_entities
get_user_dossier
list_dates

Use MCP when you want an AI agent to retrieve structured context from SCND.AI instead of asking you to paste notes manually. Treat the MCP token like a password. Regenerating it will invalidate old connections.

Analytics

Analytics is a productivity view for reviewing patterns. It currently appears as a reworked, work-in-progress section.

Available time ranges:

Today
Yesterday
Week
Month
Pick date

The screen can show:

tracked and deep-work time
focus score
productive and distraction time
daily archetype
top apps
top entities
hourly activity chart
hourly timeline
pulse cards comparing the day with a recent average
Coach Mode for AI review of a day

Use Analytics to understand time distribution and switching patterns. If a day has too little data or hourly analysis has not run yet, some areas may be empty.

Goals

Goals connects your intended work with what SCND.AI observed.

Main tabs:

Plan: add goals for a selected day.
Results: view AI-verified proof after analysis.
History: review previous goal completion.

In Plan, you can add goals manually or use Autogenerate to suggest goals. Run Analysis checks goals against Journal and memory evidence. Automatic goal checking can also run after hourly analysis if enabled in Settings.

Good goals are short, specific, and observable:

prepare proposal draft
review onboarding flow
publish landing page update
analyze support tickets

Goal history shows completion ratios by day and can expand to reveal individual goals.

Assistant

Assistant configures two awareness modes: Notice and Surface.

Notice

Notice is a hotkey-driven capture mode. It can look at the screen or listen to the room, then surface people, places, and things worth catching.

Settings include:

hotkey
only show items already in memory
entity type filters
listening language
microphone source

Surface

Surface is an ambient mode that watches the screen and floats chips for known entities.

Settings include:

on/off toggle
hotkey
optional audio listening
only-known-memory filter
entity type filters
refresh rate
repeat delay

Use Notice for a deliberate one-time capture. Use Surface when you want light ambient reminders while working.

Settings

Privacy

Privacy controls capture and data handling:

memory start mode
reminder delay
memory resolution: Light, Balanced, or Detailed
storage limit
data storage path
privacy filter
language model configuration

The privacy filter supports two approaches:

hide selected personal apps
track only selected work apps

This is one of the most important setup steps. It prevents private apps from entering Journal, Graph, Analytics, or MCP context.

Language Model settings support SCND.AI proxy mode and BYOK-style configuration for an OpenAI-compatible endpoint when enabled.

Productivity

Productivity settings include:

automatic goal checking after hourly analysis
future external sync for Outlook, Todoist, and Microsoft To Do
distraction alerts
app categories

App categories affect Analytics and Focus Score. Categories can be edited, apps can be reassigned, and categories can be marked as work activity.

Application

Application settings include:

user profile
subscription status
light, dark, or auto appearance
app version
data location

Recommended Daily Workflow

Start memory or confirm it is running.
Add 1-3 goals for the day.
Work normally.
Use Journal search when you need to recover context.
Use Memory Graph when you need a project, person, file, or tool profile.
Use MCP when an AI agent needs your work memory.
At the end of the day, review Dashboard, Analytics, open questions, and goal results.

Troubleshooting

If Dashboard looks empty, keep memory running for at least 30-60 minutes.

If a daily image is missing, the day may not have enough activity yet.

If an hour says it was not analyzed, wait for hourly analysis or refresh later.

If MCP port is disabled, turn MCP off before changing the port.

If an AI answer seems weak, check whether the relevant day has analyzed episodes and whether your Journal contains enough context.

If private content appears, update the privacy filter and review storage/model settings.

Best Use Cases

SCND.AI is most valuable for:

founders and operators who switch between many topics
developers working with AI coding agents
consultants and managers who need recall across tools
people who want a daily work log without manual note-taking
teams or individuals who rely on project memory and follow-up questions

The product is especially strong when used as a local memory layer for AI agents, not just as a productivity tracker.

Best starting pointBegin with setup, privacy, and dashboard. Then use Journal and Memory Graph once enough activity has been captured.

Agent workflowUse MCP when an AI agent needs structured work memory instead of a pasted brief.