Applied AI Verticals

Problem

Enterprise gen-AI in 2024 looked roughly the same everywhere: a pilot per industry, a bespoke RAG pipeline, a custom agent loop, a one-off UX. Smart work, none of it reusable. Every new client engagement reset the cost curve.

The hard part wasn't building any individual solution — it was that every "vertical" felt like a one-off. There was no productised path from "we have these capabilities" to "here's the financial-services product, here's the life-sciences product, here's the energy product." Sales asked for demos, demos asked for prototypes, prototypes asked for engagements, engagements rebuilt the world from scratch.

Approach

The vertical AI products sit on top of Brix (the asset marketplace) and Alloy (the engineering framework for reuse). Each vertical takes the same underlying assets — models, agents, data connectors, evaluation harnesses — and shapes them into an industry-specific package: financial services, life sciences, energy, retail.

Three decisions that mattered:

• Verticals consume from a marketplace, not a service team. Brix made the building blocks discoverable so each vertical could pull what it needed without coordination overhead.
• Industry shaping happens at the edges, not the core. Models and agents stay general; vertical-specific UX, data integrations, and workflows wrap them.
• One demo contact, one story. Instead of every engagement team telling a different version of the story, the McKinsey blog post (linked below) became the canonical pitch.

→ McKinsey · Industry AI products deliver value across a wide range of use cases

Outcome

Live across multiple industries. The McKinsey blog post now lists me as the demo contact — single point of entry for clients across all verticals — which has compressed sales cycles and reduced the "which version of this product do I show you" thrash.

Screenshot of the attribution below, in case the article gets edited upstream: