AI Launch Pipeline
Run the full AI product launch monitoring workflow in one command.
Quick Start
bash scripts/ai_launch_pipeline.sh
Output lands in $AI_PIPELINE_WORKSPACE/runs/<timestamp>/ (default: ~/.openclaw/workspace/data/ai-launch-pipeline/runs/).
Pipeline Stages
| # | Stage | Output |
|---|---|---|
| 1 | RSS monitoring | rss/rss_hits.json — AI launch hits from OpenAI, Anthropic, Google, HuggingFace, DeepSeek feeds |
| 2 | Product search | search/search_results.json — supplementary search results |
| 3 | Screenshot capture | screenshots/ — text snapshots of hit pages + manifest |
| 4 | Trend analysis | trends/summary.md + trends/trend_report.json — model/org frequency, top signals |
Configuration
AI_PIPELINE_WORKSPACE— override output root directory- Edit the
feedslist in Stage 1 to add/remove RSS sources - AI keyword filter is in the
ai_keywordsregex (Stage 1)
Requirements
- Python 3 (stdlib only — no pip packages needed)
- Optional:
duckduckgo-searchCLI for enhanced Stage 2