Agents

The three live together because they share the same internal infrastructure (orchestration, retrieval, eval) and are built as one body of practice, not as independent products.

— Equity research agent

Agent to accelerate thesis analysis, models and equity research reports. In active use on my desk: transcript reading, first-pass earnings updates, focused evidence retrieval.

Stack: LangGraph orchestration, custom retrieval over filings and transcripts, evals on proprietary datasets.

— LinkedIn writer agent

Agent to produce and sustain professional content cadence without friction. Inputs: raw notes, links, half-baked ideas. Outputs: drafts that respect voice, format and curated tone.

Stack: simple write/critique/revise pipeline, prompts iterated against my own corpus.

— Credit underwriting agent

Agent that assists credit underwriting decisions: document ingestion, policy cross-checking, recommendation drafts for human review.

Stack: structured extraction with validated schemas, recall over policy criteria, output with explicit reasoning.

Status

Active builds. Each agent tracks its own internal adoption metric; public versions ship when one crosses the "worth teaching" threshold.