Skip to main content

Integration Guides

These guides show you how to connect SynapseKit to the tools and platforms your team already uses. Each guide is self-contained — pick the one that matches your use case and you can have something running in under 30 minutes.

Guides in this section

GuideWhat you'll buildDifficultyTime
GitHub PR Review AgentAgent that fetches PR diffs and generates code review commentsIntermediate~25 min
Notion Knowledge Base with RAGQ&A system over your Notion workspace using RAGIntermediate~20 min
Slack Q&A BotSlack Bolt app that answers @mentions with RAG-powered responsesIntermediate~30 min
SQL Analytics AgentNatural language → SQL → formatted results agentIntermediate~25 min
Email Triage AgentClassify, prioritise, and draft responses to incoming emailsIntermediate~20 min
YouTube Video SummarizerExtract transcripts and generate structured summaries with timestampsBeginner~15 min
News Monitoring DashboardScheduled news monitoring with entity extraction and digest generationIntermediate~25 min

Prerequisites

All guides assume SynapseKit is installed:

pip install synapsekit

Each guide lists its own additional dependencies (e.g. slack-bolt, youtube-transcript-api) and required API keys in its Prerequisites section.

Patterns used across these guides

RAG (Retrieval-Augmented Generation) is used in the Notion, Slack, and SQL guides. The pattern: load documents → chunk → embed → index → retrieve relevant chunks at query time → pass to LLM.

FunctionCallingAgent is used in the GitHub and SQL guides. The pattern: give the LLM a set of tools (functions it can call) and let it decide which tools to invoke based on the user's request.

Loaders (NotionLoader, YouTubeLoader) are used to pull content from external sources into SynapseKit's document format.

Structured output (Pydantic models) is used in the email and news guides to parse LLM responses into typed objects for downstream processing.

Which guide should I start with?