CONSOLE/Research PipelineBlockwise

Research Pipeline

Research engine running · checked 3:54 pm

Pipeline Stages

Trigger

Cron fires, selects due locations

→

Apify Fetch

Scrapes Meta Ad Library via Apify actors

→

Ingest

Raw ads → research DB (dedup, normalize)

→

Enrich

LLM tagging, creative analysis, swipe-file scoring

→

Serve

REST API → Blockwise web (ad radar, swipe file)

Architecture

The research pipeline is owned by Blockwise — it scrapes the Meta Ad Library via Apify, ingests into a Supabase research DB, enriches with LLM tagging, and serves the ad radar + swipe file through Blockwise's REST API.

Frank Console reads the pipeline's health endpoint read-only. The pipeline-graph package renders interactive traces (seed context → resolved prompt → outputs) when a specific job is selected.

Containers: research-gateway, research-db, research-rest, research-backup, listing-scraper, steel (browser).