Northline Research Studio
Customer-facing financial research co-pilot with document grounding and compliance guardrails.
Northline surfaces a research workspace where retail and advisory users generate briefs grounded in prospectuses, filings, and approved research libraries - citations and lineage exported for audit.
Technology stack
Technology stack
- Python
- Claude
- LlamaIndex
- pgvector
- LangSmith
Representative technologies from the delivery-not an exhaustive inventory of every dependency or internal tool.
Challenge
UX demanded fast generation; compliance demanded citations, retention policies, and refusal boundaries on advice-sensitive prompts.
Approach
Python ingestion with LlamaIndex, pgvector for approved corpuses, Claude under policy prompts, LangSmith tracing, and evaluation harnesses on representative user queries.
Outcome
Placeholder programme narrative - verify before publish. Headline metrics stated for diligence review only.
Related services: generative ai solutions