# Engagements, and what they moved.

Source: https://customlabs.io/case-studies/
Updated: 2026-09-02

Case studies

# Engagements, and what they moved.

The problem we were brought in for, what we built, and the metric that changed, from real client engagements.

Client

Industry

Engagement

Headline metric

Read time

[A growth-equity firm](https://customlabs.io/case-studies/ai-diligence-flagged-replatform-cost/)

Private capital / technical diligence

Readiness & Diligence, 2 weeks

$1.4M Undisclosed re-platform cost surfaced

7 min

[A mid-market healthtech](https://customlabs.io/case-studies/extraction-notebook-to-production/)

Healthtech / clinical intake

Custom Development + AI Integrations, 8 weeks

12% → 3% Field hallucination rate

7 min

[A Series B fintech](https://customlabs.io/case-studies/model-agnostic-routing-cut-inference-spend/)

Fintech / risk & fraud

Strategy & Architecture + AI Integrations, 10 weeks

-41% Cost per resolved request

7 min

[A mid-market SaaS support platform](https://customlabs.io/case-studies/retrieval-pipeline-cut-support-load/)

B2B SaaS

AI Integrations, 8 weeks

-38% Ticket escalation rate

6 min

The client is anonymised, and these figures are not independently audited.

July 7, 2026 · 7 min read

A growth-equity firm

$1.4M Undisclosed re-platform cost surfaced

## The AI-Native Target That Wasn't: A Two-Week Diligence

A growth-equity firm's term sheet on an 'AI-native' SaaS target got repriced after a two-week check found $1.4M of hidden re-platform cost.

diligencereadinessdue-diligence[architecture](https://customlabs.io/topics/strategy-architecture/)

[Read →](https://customlabs.io/case-studies/ai-diligence-flagged-replatform-cost/)

June 30, 2026 · 7 min read

A mid-market healthtech

12% → 3% Field hallucination rate

## From Notebook to Production: An Extraction Model You Could Trust

A healthtech's extraction model dazzled in a notebook, then hallucinated on real traffic. An eval harness cut [hallucinations](https://customlabs.io/glossary/hallucination/) from 12% to 3% in 8 weeks.

[evals](https://customlabs.io/topics/evals-observability/)extraction[production](https://customlabs.io/topics/shipping-to-production/)healthtech

[Read →](https://customlabs.io/case-studies/extraction-notebook-to-production/)

June 16, 2026 · 7 min read

A Series B fintech

-41% Cost per resolved request

## Cutting Inference Spend 40% Without Betting on One Vendor

A Series B fintech's LLM spend outran revenue on one provider. Model-agnostic routing cut unit cost, vendor risk, and closed a data-residency gap.

[architecture](https://customlabs.io/topics/strategy-architecture/)[cost](https://customlabs.io/topics/ai-cost-efficiency/)[model-agnostic](https://customlabs.io/topics/strategy-architecture/)routing

[Read →](https://customlabs.io/case-studies/model-agnostic-routing-cut-inference-spend/)

May 12, 2026 · 6 min read

A mid-market SaaS support platform

-38% Ticket escalation rate

## Retrieval Pipeline That Actually Cut Support Load

A SaaS support platform swapped a keyword search widget for a tenant-isolated retrieval pipeline, cutting escalations and response time with no new headcount.

[retrieval](https://customlabs.io/topics/retrieval-rag/)supportai-integrations

[Read →](https://customlabs.io/case-studies/retrieval-pipeline-cut-support-load/)