$cd ../use-cases/
β‘ ProductivityHigh Value20 min setup
$ cat personal-research-assistant.md
await researchTopic(query, { sources: 50 })
/** I replaced my research workflow with OpenClaw. It reads 50 papers while I sleep. */
Dr. Chen is a biotech researcher who needed to review 200+ papers for a grant proposal. Manually, this would take 3 weeks. With OpenClaw's PDF Batch Mode, she dropped all papers into a folder before bed. By morning, she had a 15-page structured literature review with themes, contradictions, and citation-ready key findings. Total processing time: 47 minutes. She submitted the strongest proposal of her career β and won the $2M grant. Now she runs a weekly 'Research Pulse' that monitors 30 journals and competitor lab publications automatically.
research_modes.md
π¬ Research Modes
Web Research Mode
Tell OpenClaw a topic. It searches the web, visits 20-50 sources, extracts key facts, and compiles a structured briefing with citations.
Research the current state of AI regulation in the EU. Focus on 2025-2026 developments.
PDF Batch Mode
Drop 10-100 PDFs into a watched folder. OpenClaw reads them all, extracts themes, finds contradictions, and generates a literature review.
$ cp ~/Downloads/papers/*.pdf ~/openclaw/research/inbox/
Competitive Intelligence Mode
Monitor competitors' blogs, job postings, press releases. Weekly digest of what your competitors shipped.
Every Monday, summarize what Notion, Linear, and Coda shipped last week.
config.yaml
βοΈ Configuration
# openclaw/config.yaml
skills:
research:
enabled: true
web_search: true
max_sources: 50
output_format: "markdown"
briefing_example.md
π Example Briefing Output
# AI Regulation EU β Research Briefing
Generated: 2026-03-01 08:00 | Sources: 43 | Processing: 6 min
## Executive Summary
The EU AI Act entered full enforcement in 2025. High-risk systems face strict conformity assessments.
## Key Findings (7)
1. Prohibited AI practices enforced since Feb 2025
2. High-risk classification expanded in Oct 2025
3. SME exemptions narrowed
2. High-risk classification expanded in Oct 2025
3. SME exemptions narrowed
β FAQ
Q1. Can it read academic PDFs with complex layouts?
Yes. OpenClaw uses vision-capable LLMs to parse multi-column layouts, tables, figures, and citations. It handles arXiv, Nature, and Elsevier formats reliably.
Q2. How many sources can it process at once?
Web mode: up to 50 sources per query. PDF batch: tested with 200+ files. Processing scales linearly β 50 PDFs take ~30 minutes with local models, ~10 minutes via API.
Q3. Does it hallucinate or make up sources?
Unlike ChatGPT, OpenClaw cites the actual source URL or PDF filename. You can verify every claim. Hallucination rates are 95% lower because it's extracting, not imagining.
Q4. Can it monitor topics automatically?
Yes. Set up a cron schedule and OpenClaw will run web research weekly, comparing new findings against your last briefing and highlighting what's changed.
Q5. Is my research data private?
Completely. PDFs are processed locally. With local LLMs, zero data leaves your machine. Even with API models, papers are processed in chunks β not sent in full.