Pod

Share
Research Architecture

How the Pod Works

An AI-native buyside research pod focused on listed Japanese consumer and consumer-adjacent equities, supervised by a former hedge fund analyst.

Human-supervised Bottom-up research Macro + valuation + alt data Priority-driven workflow
Organization

Built like a research operating system

The pod is organized around a simple principle: one final human decision-maker, one central orchestrator, and a group of specialist AI analysts working across coverage, valuation, macro, newsflow, alternative data and risk.

Human Supervision

Former Hedge Fund Analyst

Final judgment, thesis approval, prioritization and editorial control.

Central Coordination

PM Orchestrator

Routes work, assigns priorities, escalates catalysts and maintains research memory.

Coverage

Coverage Captains

Primary owners of single-name work. They maintain the business model, competitive position and core thesis.

Valuation

Valuation Agent

Tracks expectations, peer sets, scenario analysis and valuation dislocations.

Macro

Macro Agent

Monitors wages, inflation, demand conditions, tourism, pricing power and the consumer cycle.

Catalysts

Newsflow Agent

Tracks earnings, guidance changes, governance events, buybacks and other company catalysts.

Channel Work

Alternative Data Agent

Looks for supporting or conflicting signals outside traditional disclosures.

Risk

Red Team Agent

Pressure-tests assumptions, challenges the thesis and highlights downside risks.

Workflow

How research is produced

The workflow is event-driven and priority-based. The system monitors the coverage universe continuously, activates the relevant specialists when something matters, and publishes only reviewed work.

01

Monitor

Track companies, sectors, macro signals and catalysts across the coverage universe.

02

Detect

Identify events that matter, from earnings and guidance changes to valuation dislocations and macro shifts.

03

Debate

Gather focused input from specialist agents across coverage, valuation, macro, newsflow and risk.

04

Review

Consolidate the work under human supervision and challenge weak assumptions before publication.

05

Publish

Release only reviewed outputs through the Coverage Map, Pipeline, Consumer Pulse, Newsflow Monitor and Deep Dives.

Technology Stack

Research stack

Communication Layer

Structured Workspace

Threaded research discussions, alerts and workflow visibility.

Orchestration Layer

Workflow Engine

Activates the right agents at the right time and routes work across the pod.

Memory Layer

Research Memory

Stores company files, research history, thesis changes and catalyst logs in structured form.

Data Layer

Market + Company + Alternative Data

Combines public disclosures, market data, newsflow and non-traditional signals.

Outputs

What the pod produces

Coverage Map

The investable universe and its internal classification.

Pipeline

Priority ranking across top ideas, watchlist names and non-actionable coverage.

Consumer Pulse

A macro view of the Japanese consumer environment.

Newsflow Monitor

Event tracking across companies, sectors and catalysts.

Deep Dives

Long-form underwriting on selected names.

Objective

Designed for clarity, speed and discipline

The goal is simple: combine breadth, analytical discipline and speed in a research process built for listed Japanese consumer equities.