Skip to content

// Work

AI case studies: shipped systems, measured outcomes

Selected work, anonymised for confidentiality. The numbers are the point — the outcome, the metrics that moved, and the stack we shipped.

// Featured

Fintech

Document intelligence

Analysts hand-reviewed every filing — dense, inconsistent, slow — before anything downstream could move.

We shipped an extraction-and-review pipeline that reads dense financial documents, surfaces only what needs a human, and routes the rest straight through. Grounded, traced, and gated on an eval suite — so accuracy is measured, not assumed.

  • LangGraph
  • pgvector
  • Postgres
  • Go
Read the case studyCapability: LLM Integration & Evals
-82%

manual review

vs. the fully-manual baseline

throughput per analyst

3.5×

vs. the pre-automation baseline

// Track record

15+systems shipped
6+ yrsin production
~4 wksto a first slice

Client names are under NDA — on a call we’ll walk you through the architecture and share references.

// More work

More selected work

Healthtech

Clinical RAG

Clinicians needed answers trustworthy to the citation, not plausible summaries.

  • pgvector
  • LlamaIndex
  • OpenAI
Capability: RAG & Knowledge Systems
0hallucinated citations in eval

100% · source-linked answers

E-commerce

Support agent

A support queue growing faster than the team could hire.

  • LangGraph
  • Anthropic
  • River
Capability: AI Agents & Automation
47%tickets auto-resolved

-38% · first-response time

Logistics

Demand forecasting

Replenishment ran on gut feel and stale spreadsheets.

  • Python
  • Postgres
  • Docker
-31%stockouts

+12% · forecast accuracy

Media & Streaming

Moderation agent

A moderation backlog that grew with every upload spike.

  • LangGraph
  • OpenAI
  • River
-91%time-to-decision

6× · reviewer throughput

Insurance

Claims triage

First-notice-of-loss claims piled up in a shared inbox before anyone could route them.

  • LangGraph
  • Anthropic
  • Postgres
  • River
Capability: AI Agents & Automation
-64%time to first action

3× · adjuster capacity

Manufacturing

Predictive maintenance

Line stoppages were discovered when a machine failed, not before.

  • Python
  • Postgres
  • Docker
  • Grafana
Capability: MLOps & Infra
-43%unplanned downtime

12 days · median early warning

// Let's build

Want a number like these on your problem?

Tell us where you are. We reply within a day with a concrete next step.