Skip to content
Back to home

Documented cases

Results with the source attached

Public enterprise deployments, not Amalgama client claims. Every figure links to its original source.

01 / Customer Support

AI agents now handle roughly half of all support interactions. The support team was rebalanced from about 9,000 to about 5,000 people, and support costs fell about 17%.

Sector

Global enterprise software · customer support

Challenge

A high volume of routine support questions on the company’s help portal created queues and slow resolution for common issues.

Solution

Autonomous AI support agents resolve common questions end-to-end and hand off complex or sensitive cases to human reps.

Systems and timing

Help portal · CRM · knowledge base. Rolled out through 2025.

Salesforce (Benioff) / CNBC, 2025
02 / Research & Intelligence

Cut customer scam losses by ~50% and customer-reported fraud by ~30%, while AI-assisted app messaging reduced call-centre wait times by ~40%.

Sector

Major retail bank · fraud & risk monitoring

Challenge

Scams and fraudulent transactions were hard to catch in real time across millions of daily payments.

Solution

Generative-AI monitoring flags suspicious transactions and scam patterns as they happen and warns customers before money leaves the account.

Systems and timing

Real-time transaction monitoring · customer alerts. Reported 2025.

Commonwealth Bank, 2025
03 / Internal Knowledge

Trained on documentation from nine in-house lawyers, it now helps answer more than 40,000 client legal questions a year. Many receive an answer in under 24 hours instead of days.

Sector

Global bank · legal & compliance

Challenge

Client-facing teams waited on a small legal department for routine questions, stretching answers out to days.

Solution

A legal-assistant GPT, trained on the bank’s own legal documentation, answers routine questions from approved sources and escalates novel matters to lawyers.

Systems and timing

Internal legal knowledge base · ChatGPT Enterprise. 2025.

BBVA / OpenAI, 2025

Start with one workflow worth fixing.

A focused assessment identifies the strongest use case, the constraints, and the practical next step.