The AI Pragmatist's Manifesto
The Newcastle Network’s AI Maturity Model outlines how we harness AI’s potential, scale processes and decision making, and prepare for what’s next.
We take a pragmatic approach to AI: Embrace the technology at the speed it’s available, even if it’s not fully developed.
This mindset allows our team to harness its potential today while preparing for its full capabilities tomorrow.
Earlier in my career, I witnessed a similar inflection point during the dot-com era — transformative time that spawned some of the most innovative ideas and enterprises. AI is poised to make an even greater impact.
At The Newcastle Network, AI integration is central to our strategy. From day one, we built our business with data science and technology at its core. Over the past 20 months, we’ve taken deliberate steps to embed AI into our processes and decision-making frameworks.
Introducing our AI Maturity Model
The Newcastle Network’s AI Maturity Model
Borrowing from capability-maturity models that technology organizations have used for years, we created our own AI Maturity Model — a simple framework to grade our level of AI integration across our organization and operating processes. The levels are as follows:
1. Human-centric
These are processes that leverage the old-fashioned meat computer. While many of our activities start in this category, we do expect a number of them to remain here even as AI technology evolves.
Examples: sourcing strategies, underwriting model development
2. Human-led, AI-assisted
Today, this represents the easiest way to capture value from AI as familiar technology tools introduce incremental AI capabilities to improve productivity. At Newcastle, we have taken this one step further, building our own custom stack of AI agents that can use our proprietary data and analytic modules in unique ways to support us.
Examples: investment theme development, performance benchmarking
3. AI-led, human-assisted
More sophisticated AI capabilities can work independently but require “humans-in-the-loop” to ensure directional integrity and deliverable quality. Often, these activities have a time-oriented aspect to them, like compiling and curating a collection of relevant facts on a topic. We use our current agentic AI stack to perform semi-autonomous tasks that notify us when input or review is required.
Examples: portfolio monitoring (public sources), information ingestion and extraction
4. Fully AI-driven
We expect that some of our processes — after benefiting from reinforcement learning from human feedback — will have the potential to go full AI. Our initial plan is to replace some onerous but important last-mile activities, like generating memos from previously vetted facts and data, and we have prototypes in the works.
Given today’s technical capabilities, we can’t imagine going much further than this (at least that’s what my AI overlords have instructed me to say…)
To effectively deploy our framework, we needed an inventory of all of our processes, which was an effort we had already undertaken. Now we have a roadmap outlining both where we are and where we want to go over time. Further, it has been a great tool to help prioritize “what’s next” as we think about ways to improve the productivity and impact of our team.
Every firm, regardless of industry, can apply this playbook. And it can start at any level of your organization. The key is to begin, learn and stay adaptable.
AI isn’t just a tool — it’s an opportunity to rethink how we collectively create value.
Chris Casgar,
Managing Partner
As Managing Partner, Chris oversees all aspects of the firm, including, most importantly, the development of our people and the execution of our strategy. View profile.
