Scout

Company intelligence agent that navigates your workspace, builds context, and answers questions.

Scout is a company intelligence agent that gets smarter as it explores your systems.

Most AI assistants require you to upload documents or paste context. Scout navigates to the source: it connects to your Slack, Google Drive, databases, and wikis, then builds a knowledge graph of your company as it works.

How It Works

Scout uses Context Providers to connect to your systems and retrieve information on demand:

Provider	What It Accesses
Web	Public URLs, documentation sites
Slack	Channel history, threads, search
Google Drive	Documents, spreadsheets, presentations
Database	CRM records, product data
Wiki	Internal documentation (Git-backed or Notion)
MCP	Any tool exposed via Model Context Protocol

Scout doesn't just search. It navigates: following links between systems, resolving references, and building a map of how information connects.

Self-Building Knowledge

As Scout answers questions, it maintains two internal systems. See Knowledge & CRM for details.

System	Purpose
Wiki	Curated facts about your company: org structure, product names, key metrics
CRM	People and relationships: who works on what, who to ask about X

These start empty and grow with use. Scout learns that "the API" means your REST API, not a generic concept. It learns that Sarah owns billing and Marcus handles infrastructure.

Run Locally

See Setup guide for detailed instructions.

1git clone https://github.com/kern-agi/scout.git && cd scout
2
3cp example.env .env
4# Edit .env: add OPENAI_API_KEY and configure providers
5
6docker compose up -d --build

Confirm Scout is running at http://localhost:8000/docs.

Connect to the Control Plane

Open os.kern.ndx.rocks
Click Connect OS → Local
Enter http://localhost:8000

Deploy to Railway

See Deploy to Railway for JWT setup and production configuration.

1railway login
2./scripts/railway_up.sh

The script provisions PostgreSQL, configures environment variables, and deploys Scout.

Then connect via the control plane:

Open os.kern.ndx.rocks
Click Connect OS → Live
Enter your Railway domain

Configure Providers

Scout's power comes from the providers you connect. Start with one or two, then add more as needed.

Slack

See Slack setup for bot creation and permissions.

1# Add to .env
2SLACK_BOT_TOKEN=xoxb-...
3SLACK_SIGNING_SECRET=...

Google Drive

See Google Drive setup for service account configuration.

1# Add to .env
2GOOGLE_SERVICE_ACCOUNT_FILE=/path/to/service-account.json

MCP Servers

See MCP setup for server configuration. Connect any MCP-compatible tool:

1from kern.tools.mcp import MCPTools
2
3mcp_tools = MCPTools(servers=["npx -y @anthropic/mcp-server-filesystem"])

Example Prompts

Try these after connecting your providers:

What's the status of Project Phoenix?
Who should I talk to about billing issues?
Summarize last week's engineering discussions
What decisions were made in the Q3 planning doc?

Source

For setup guides, provider configuration, and architecture details, see the GitHub repo.