What challenges did Junson Capital face before AI implementation?

Extract structured intelligence from high-variance, unstructured documents. Reduce the 15.5 days required to generate a single report. Automate manual internal reimbursement and invoice matching workflows.

What was the impact of the AI solution?

Reduced report generation time from 15.5 days to 1 day. Reduced monthly internal reimbursement workflows from 10 days to 1 hour. Achieved 90%+ resource efficiency in data extraction. Enabled 2 people to handle work that previously required 15.

Source

GCPProductionEvidence: Medium65/100

Junson Capital automates global investment reporting with Vertex AI, Gemini and Document AI

Use case typeIntelligent document processingUpdated Jun 13, 2026

Junson Capital is a global investment management firm that built an AI-native data engine on Google Cloud to extract structured intelligence from high-variance, unstructured documents. The firm uses Vertex AI, Document AI, and Gemini models to process private bank statements, inspection testing reports, and handwritten documents with a proprietary six-layer recognition framework. The implementation also supports regional hosting for data residency and powers an AI-native digital employee workflow for operational tasks.

Organization: Junson Capital
Industry: Finance
Location: Hong Kong
Published: May 2026

Reported outcomes

−93.5%

report generation timeTime & speed

−99.6%monthly reimbursement workflow time90%productivity

Strategic outcomes

New product / capabilityBuilt AI-native document extraction engineSpeed & agilityAutomated report generation workflowSpeed & agilityAutomated reimbursement and invoice workflowsRisk & complianceEnabled regional data residency hosting

Why do we believe this?Outcome claims, sources, and evidence checks

Normalized claim

Report generation time: 93.5% decrease

Google Cloud Customer StoryMay 26, 2026Customer storyExplicit claimMedium evidence strength

Reduced report generation time from 15.5 days to 1 day.

Correction recorded Jul 26, 2026 · Previously: 15.5 days decrease · Replaced endpoint-only extraction with one source-verified duration transition.

Normalized claim

Monthly reimbursement workflow time: 99.6% decrease

Google Cloud Customer StoryMay 26, 2026Customer storyExplicit claimMedium evidence strength

Reduced monthly internal reimbursement workflows from 10 days to 1 hour.

Correction recorded Jul 26, 2026 · Previously: 1 days decrease · Removed a duplicate endpoint after preserving the full duration transition.

Normalized claim

Productivity: 90%

Google Cloud Customer StoryMay 26, 2026Customer storyInferred claimMedium evidence strength

Achieved 90%+ resource efficiency in data extraction.

Correction recorded Jul 26, 2026 · Previously: 10 days decrease · Replaced endpoint-only extraction with one source-verified duration transition.

Why do we believe this deployment?Customer identity, provider attribution, maturity, and source checks

Customer: Junson Capital
Provider: GCP
Maturity: Production
Linked source: Google Cloud Customer Story

The implementation also supports regional hosting for data residency and powers an AI-native digital employee workflow for operational tasks

Customer identity supportedSource describes one deploymentMaturity supported

Primary read

Use case focus

Showing 3 of 4

1Document Processing
2Workflow Automation
3Knowledge Extraction

Extract structured intelligence from high-variance, unstructured documents.
Reduce the 15.5 days required to generate a single report.
Automate manual internal reimbursement and invoice matching workflows.

Built a proprietary six-layer recognition framework using Vertex AI and Document AI for document extraction.
Used Gemini 2.5 Flash and Gemini 2.5 Pro for document understanding and hosted Gemini in Google Cloud regions for residency.
Created an AI-native 'digital employee' workflow that integrates extracted data into operational processes.

Technologies

Vertex AI Document AI Gemini 2.5 Flash Gemini 2.5 Pro Google Cloud Regions

Reduced report generation time from 15.5 days to 1 day.
Reduced monthly internal reimbursement workflows from 10 days to 1 hour.
Achieved 90%+ resource efficiency in data extraction.
Enabled 2 people to handle work that previously required 15.

Architecture

Junson Capital built a proprietary six-layer recognition framework on Google Cloud that combines Vertex AI and Document AI for structured document extraction, with Gemini 2.5 Flash and Gemini 2.5 Pro used for document understanding. The solution uses Google Cloud regions for residency and powers an AI-native digital employee workflow for operational processing.

Sources & evidence1

Evidence: Medium65/100Evidence strength