The company story
A global hedge fund with roughly $26Bn AUM needed to check investment theses faster and across more sources without losing rigor. Through New Value Solutions, we helped design and deliver the agentic research system.
Research cycle focus
Days → hours
Cycle
Days → hours
Architecture
Multi-agent
Context
$26Bn AUM
A $26Bn Hedge Fund — impact snapshot
The pressure
What was on the line
Research cycles were constrained by time and source coverage. Checking a thesis had to go from days to hours without lowering the quality analysts would trust.
What changed
The arc of the work
- 01
Decompose the research problem
Investment theses were broken into validation paths an agentic system could run — so coverage expanded and analysts never had to learn to write prompts.
- 02
Orchestrate agents and evidence
A tools-first multi-agent research architecture (including GPT-4, Claude, and Gemini) with traceable reasoning — so analysts could see how a thesis was checked, not just a score.
- 03
Return time to judgment
Analysts spend less time collecting and more time deciding. Research cycles went from days to hours.
Before and after
What people could feel in the week
Checking a thesis took days
→Checking a thesis takes hours
Source coverage limited by human bandwidth
→Agentic loops expand what analysts can check
Where they are now
The outcome that stuck
The research team keeps the judgment; the system does the collecting and cross-checking. It runs on Azure in a regulated environment, delivered through New Value Solutions.
For technical readers
How it was built
Multi-model architecture (GPT-4, Claude, Gemini), multi-agent research orchestration with LangChain/LangGraph, thesis decomposition and validation, FastAPI, Azure, PostgreSQL, Redis. Delivered through New Value Solutions.
LangGraph · GPT-4 / Claude / Gemini · FastAPI · Azure · PostgreSQL · Redis
More company stories
- CopperTeamsCopperTeams’ 20% team-performance guarantee — extended with AI coaching between human sessions.
- Athyna InsightExpert-validated building code answers in seconds — with licensed professionals in the loop.
- A Canadian Government AgencySiloed housing archives became searchable — with high-confidence document classification in a two-phase POC.