Research Agent

Deep Research, Automatically Delivered Every Morning

Staying informed takes time you don't have. Reading industry news, tracking competitor moves, following regulatory changes, and monitoring emerging topics — it's all valuable, but the work of gathering and synthesizing it is pure overhead. The Research Agent does that work for you: deep daily research on your configured topics, synthesized into actionable briefings, delivered to Telegram every morning.

This blueprint configures an OpenClaw agent with a rigorous research mindset — cross-referencing sources, flagging uncertainty, and always citing links — rather than generating plausible-sounding summaries from training data.

What the Agent Does

Daily Morning Digest (7 AM): For each topic configured in your USER.md, the agent searches for new developments, fetches and reads full articles from credible sources, cross-references multiple results, and compiles a structured briefing. Each topic section leads with the key finding, includes source links, flags conflicting reports or uncertain information, and notes the "so what" — why it matters for you. The briefing arrives in Telegram before your day starts.

Ongoing Topic Tracking: The agent maintains a memory/topics.md file with ongoing research threads: open questions, promising sources, threads to follow up. This means research compounds over time — the agent builds context on your topics across weeks and months, not just searching fresh each day.

Deep Dives (on demand): When you need more than a daily briefing, you can ask the agent for a thorough investigation of a specific question. It will fetch multiple sources, read full articles, synthesize conflicting perspectives, and deliver a comprehensive summary with citations.

Weekly Summary (Friday): A synthesized summary of everything gathered across the week — key themes, emerging patterns, and open questions for the following week.

The agent's SOUL.md instills a specific research discipline: never hallucinate sources, always prefer primary sources, cross-reference before reporting, flag uncertainty explicitly ("unclear," "conflicting reports"), and compress 20 sources into 5 actionable bullets.

Who This Is For

Founders, analysts, investors, journalists, policy researchers, and anyone who needs to stay current across multiple domains without spending hours reading every day. Especially effective for people tracking fast-moving fields where missing a development has real consequences.

What's Included

Setup Steps

1. Copy the blueprint into ~/.openclaw/workspace/
2. Edit USER.md — list 3–8 research topics, specify any preferred sources or outlets, set your timezone
3. Connect Notion (optional) — if you want summaries saved with source links for future reference
4. Configure Telegram for briefing delivery
5. Start the agent — openclaw start; your first briefing will arrive at 7 AM tomorrow

What Rigorous Research Means Here

Most AI research tools summarize from training data — they tell you what was true when the model was trained. This agent searches the actual web, fetches and reads actual articles, cross-references multiple sources, and flags when sources disagree. If it can't find credible information on something, it says so rather than generating a plausible-sounding non-answer.

Why This Beats Reading the News Yourself

Reading and synthesizing research manually takes 45–90 minutes per day for someone tracking multiple topics seriously. With this agent, that synthesis happens automatically and arrives as a tight, actionable briefing. You read for 10 minutes instead of 90, and you don't miss the buried story that matters.

The agent gets better over time — as it builds context on your topics in memory/topics.md, its briefings become more targeted and its pattern recognition improves. Week 4 is meaningfully better than Week 1.