Core Design Principles

Effective agents follow consistent design patterns regardless of whether they operate in Google Workspace, Notion, Slack, or any of the hundreds of available integrations. These principles ensure your agents remain reliable, maintainable, and scalable across different platforms.

TLDR: Design agents with single, clear purposes. 
Define specific boundaries for what each agent handles.
Keep tasks focused rather than trying to build one agent that does everything

Single Responsibility Principle

Each agent should have one clear, well-defined purpose. This principle applies whether your agent manages calendars, processes data, or coordinates communication across multiple platforms.

Good Examples

• "Calendar Analytics Agent" - Analyzes meeting patterns and generates reports

• "Document Sync Agent" - Keeps project status updated between Notion and Slack

• "Lead Qualification Agent" - Processes new contacts from Apollo.io and updates CRM

Avoid These Patterns

• "Everything Manager Agent" - Handles calendar, email, documents, and reporting

• "Multi-Purpose Assistant" - Does whatever the user asks for

Clear Input and Output Boundaries

Define exactly what data your agent expects to receive and what it will produce. This is especially important when working across different app ecosystems with varying data formats.

Input Specifications

• Expected data sources (specific calendars, Notion databases, Slack channels)

• Required parameters (date ranges, user filters, project identifiers)

Output Specifications

• Target destinations (Google Sheets, Notion pages, Slack notifications)

• Format requirements (tables, charts, formatted text, structured data)

• Success criteria (completion confirmations, error handling, status updates)

Universal Pattern To Follow

• Data collection → Processing → Output generation