Artificial Intelligence Engineering Maturity Model

Strong engineering principles are at the core of scalable and effective artificial intelligence products and services. This maturity model combines principles from our extensive experience, as well as some elements of other similar public models.

Early Interest: Consideration of AI potential - Maturity Stage

At this maturity stage, a business is potentially in a mixture of states of mind. Some teams may be overly excited, and others cautious.

Problems

Losing market share to more nimble competitors

Legacy inertia

Painful manual insight generation

Expert resources spending time on manual, static reporting processes

Hampered by slow, expensive, and painful innovation

Capabilites

A working environment that fosters innovation

Experimentation: Hopeful in AI and its promise - Maturity Stage

At the AI Experimentation maturity stage, a business is "learning through doing" and trying new approaches. At this stage a business must work hard to avoid distractions and fads.

Problems

Innovation is slow and expensive

Data Scientist resource wasted on non-differentiating features

Distraction caused by technology fads

Capabilites

Media assets include structured, descriptive metadata

A working environment that fosters innovation

A safe environment to prototype new ideas using real data

Operational: AI in Production with successes and challenges - Maturity Stage

A business has successfully moved an important AI capability into production at this maturity stage. The service is working, but might be more expensive than planned, and there are unforeseen challenges.

Problems

Lack of engineering best practices for AI

Lack of automation for testing and deploying AI

Lack of governance and change management for AI

Struggling to run AI in Production

Emerging: AI Data Science and Ops Capabilities - Maturity Stage

At this maturity stage, AI is pervasively used for some business models, and there is an emerging data science and AI ops capability. There are still likely to be pain points around ML training overheads process inefficiencies.

Problems

Expensive AI training overhead

Data Scientist resource wasted on non-differentiating features

Surprisingly high Total Cost of Ownership (TCO) for Data Platforms

Transformational: AI is part of Business DNA - Maturity Stage

At this maturity stage, AI is part of the business DNA, and is running in harmony as part of the software delivery culture. AI services delivery, alongside all platform engineering ops, are operating in a productive rhythm, and the business is investing only in core differentiators.

Capabilites

The ability to move AI from R&D into production with ease

Feedback loop for AI training corpus

AI does the heavy-lifting for subject matter experts