Case studies

Real systems. Real outcomes.

A selection of recent engagements across data platforms, AI products and software licensing. Names withheld; numbers verified.

Insurance

Lakehouse migration & GenAI assistant for a global insurer

Challenge

A sprawling on-prem warehouse and three overlapping vendor contracts were slowing releases and ballooning costs. Underwriters spent hours hunting for policy precedents.

Approach

We re-architected onto a unified lakehouse, consolidated licensing across vendors, and shipped a RAG-based assistant grounded in 12 years of underwriting decisions.

42%: Lower platform cost
90 d: GenAI to production
1,200: Daily users
3→1: Vendors consolidated

Retail

Real-time personalization for a multi-brand retailer

Challenge

Batch recommendation jobs were 24 hours stale, hurting conversion on flash sales. Marketing couldn't experiment without engineering tickets.

Approach

We built a streaming feature store and a self-service experimentation layer, plus a managed model serving stack with full evaluation and rollback.

+18%: Conversion lift
<200ms: Recommendation latency
5×: Experiments shipped / week
11: Brands onboarded

B2B Software

License consolidation saving 40% for a scale-up

Challenge

A fast-growing SaaS had accumulated overlapping data, AI and observability tools through years of bottom-up adoption. Renewals were chaotic and forecasting impossible.

Approach

We ran a six-week software estate review, renegotiated four core contracts at partner-tier pricing, and stood up a lightweight FinOps practice for ongoing optimization.

40%: Annual savings
$2.1M: Year-one impact
9: Tools rationalized
6 wk: Time to close

Want to talk through a similar problem?

We're happy to share more detail under NDA, including references.

Start a conversation