Question quality
Narrow prompts around an evidence test, not a vague request to find good stocks.
SHARK MONEYYACHT CLUBSearch Shark Money
Search public and private companies, people, funds, markets, tools, and research.
Shark AI / Research agentsBuild a research process
Build a research process
that remembers every run.
Configure companies, questions, evidence lanes, and preferred cadence. Run a live multi-company scan and preserve each result in your private research history.
Agent fleet
0 agentsYour active monitors
No research agents yet
Create the first monitor to turn a watchlist into a repeatable process.
Research memory
0 recordsSaved intelligence runs
No saved runs yet
Run an agent once and its evidence snapshot will appear here for comparison.
Agent operating standard
Automate the process—not the conviction.
A strong research agent repeats a bounded question, preserves sources, exposes uncertainty, compares changes over time, and hands judgment back to the member.
Source coverage
Track which primary and secondary sources were checked, unavailable, or stale.
Change detection
Prioritize what changed since the last run instead of rewriting the entire company.
Decision relevance
Every finding should connect to a thesis assumption, risk limit, or follow-up action.
Confident synthesis
Fluent language can hide missing sources, entity errors, or unsupported causality.
Scope drift
Broad agents accumulate irrelevant findings and become difficult to verify.
Untracked revisions
Without run history, the research process cannot show what changed or why.
Define the claim
Write the exact assumption the agent is testing.
Declare sources
Specify acceptable evidence, freshness, exclusions, and failure behavior.
Require citations
Preserve direct links, dates, excerpts, and entity identifiers.
Human sign-off
Review the evidence and countercase before changing a decision.