A comprehensive research brief citing timestamped evidence from YouTube and podcast transcripts to support or refute investment hypotheses.
The agent receives a specific research question and uses the Arcmira MCP server to search for relevant mentions in public conversations. It performs multiple searches to find supporting and contradicting evidence, compares sources, and verifies claims against available data. The agent then synthesizes these findings into a structured research brief, ensuring all assertions are backed by cited transcript moments.
- Category: Research and knowledge
- Effect: Read-only
- MCP servers: Arcmira: YouTube Transcript Search (https://koskamo.com/explore/arcmira-mcp)
- Source: Investment research with YouTube video and podcast transcript search - Arcmira API (https://arcmira.com/docs/workflows/investment-research)
- Source quote: "Search company mentions, executive interviews, and expert commentary in YouTube and podcast transcripts. Compare evidence and cite original moments in investment research."
- Playbook page: https://koskamo.com/explore/use-cases/conduct-investment-research-using-transcript-search
Prompt: Investigate whether there is a material reason for the recent semiconductor selloff despite strong AI demand. Use the last 60 days. First verify the timing and size of the selloff. If the premise does not hold, say so and identify what changed. Search public conversations about NVIDIA, TSMC, ASML, Micron, Broadcom, and their GPU-cloud customers. Focus on demand, cancellations, customer agreements, lead times, power availability, margins, and financing. Separate current conditions from risks to future expansion. Find the strongest evidence supporting and contradicting each explanation, identify which companies are exposed, and cite the original moments. Finish with the unresolved questions and indicators that would change the conclusion. Verify material claims against filings and earnings disclosures.