Research Pipeline
Discovers content ideas, SEO quick wins, and trending topics through multi-source research.
Research Pipeline
The Research Pipeline is a multi-step workflow that discovers content ideas, SEO quick wins, and trending topics by combining web research, social listening, and internal performance data.
Input Schema
{
date: string; // Research date (ISO format)
type?: 'daily' | 'seo-audit'; // Research type
}Pipeline Steps
Step 1: Define Research Angles
The Researcher agent analyzes the brand profile and past performance to define multiple research angles, each with specific search queries:
{
title: string; // Research angle name
queries: string[]; // Search queries for this angle
}Step 2: Execute Research
For each angle, the Researcher runs parallel searches:
- Exa Web Search — 3–5 queries per angle for web intelligence
- X/Twitter Search — Social sentiment and trending conversations
- Web Extract — Full article extraction for key sources
- RAG Queries — Past content and performance data from semantic memory
Step 3: Score & Rank Ideas
Each discovered idea is scored and structured:
{
topic: string;
keyword: string;
searchVolume?: number;
keywordDifficulty?: number;
confidence: number; // 1–10
channels: ('blog' | 'instagram' | 'linkedin' | 'x' | 'email' | 'facebook')[];
blogAngle?: string;
socialHook?: string;
emailAngle?: string;
reasoning: string;
}Step 4: Identify Quick Wins
Finds SEO quick wins — keywords where you're already ranking positions 5–20 with decent volume that could be pushed to page 1 with targeted content:
{
keyword: string;
currentPosition: number;
volume: number;
}Output Schema
{
ideas: Idea[]; // Scored content ideas
seoQuickWins?: QuickWin[]; // Keywords close to page 1
trendingTopics?: string[]; // Trending topics in your niche
competitorGaps?: string[]; // Content gaps vs. competitors
}Scheduling
Runs daily via Inngest cron. The daily type focuses on fresh opportunities; the seo-audit type does a deeper analysis of existing ranking performance.