By now, most businesses have come to recognize the value that AI brings to enterprise transformation projects (such as those that focus on modernizing enterprise software platforms). When used properly, AI can dramatically accelerate a variety of tasks across all stages of the process, from design and implementation to testing and validation, to post-transformation management.

That said, just because businesses are leveraging AI to accelerate transformation doesn’t mean they’re all using it to maximum effect. On the contrary, my experience in the field helping to guide transformations reveals that some organizations are much more mature than others when it comes to how they take advantage of AI.

The Three “Operating Modes” of AI-Enabled Transformation

Specifically, I’ve found that businesses operate within one of three distinct modes regarding AI-enabled transformation.

1. Chatbot-Assisted Modernization

The first and most basic level involves using chatbots to assess transformation needs, formulate modernization plans and, in some cases, implement new code.

Chatbots are relatively easy to use and implement, and many teams are quickly wowed by them because they can answer questions that traditionally took a long time to research.

But chatbots are also limited in their transformation capabilities. Many lack the controls necessary to guarantee specific outcomes, which can be deeply problematic when dealing with enterprise systems that need to be modernized in highly reliable and predictable ways. They’re also only as good as the data driving them, and most chatbot implementations are not accompanied by advanced data context solutions (like knowledge graphs) that supercharge their capabilities.

2. Deliverable-Focused AI

The middle tier of AI-assisted transformation centers on what I call deliverable-focused AI. I’m referring to a strategy that treats transformation projects as a network of deliverables (such as modernizing specific code or redesigning a particular workflow), and uses AI models and agents to implement them one-by-one.

Another key characteristic of strategies at this level is the deployment of controls that can help standardize deliverables. For instance, a transformation team might take advantage of prompt markup language to help structure AI prompts and make results more deterministic than those from a generic chatbot.

This approach is powerful, but it has its limitations. Chief among them is that it treats deliverables as individual objects, making it difficult to repeat and scale transformation workflows.

3. AI Operations Platform

The most mature level of transformation involves using an AI operations platform. This means a solution that injects AI into every stage of a transformation project in an end-to-end fashion. It breaks the stages into discrete deliverables and leverages techniques like workflow templates to ensure that the transformation process is consistent, repeatable and scalable.

At its core, an AI operations platform is composed of the same underlying capabilities and technologies (like AI models, agents and prompt markup) as less mature transformation tiers. The difference is that with an AI platform, businesses integrate everything into a centralized solution, rather than using siloed AI tools and processes to address various parts of the transformation process.

The benefits of this approach extend beyond added efficiency. They’re also about generating evidence and ensuring traceability, which is precisely what businesses need to align with governance goals and meet auditors’ expectations.

After all, saying that your modernized code is compliant because you trust the AI chatbot that created it (to take one example of an outcome you might have if you are at the chatbot-centric stage of AI-driven transformation) is not something most auditors would find compelling. It’s a different story if you can identify and explain the specific prompts, prompt controls, models, agents and tools that produced and validated the code.

Advancing on the AI-Assisted Transformation Journey

Obviously, the place businesses should be at on the maturity scale I’ve described above is the third tier, meaning they should implement an AI operations platform to drive their transformation journey. Most, however, are stuck at the chatbot stage.

How can they move upward? The answer centers mostly on connecting and integrating AI tools and data sources. Deploying a chatbot is easy enough (which is why that’s where many transformation strategies start). What’s harder are tasks like constructing knowledge graphs, generating custom prompt markup code and defining deliverable standards.

These are not challenges that tools alone can solve. They require real engineering expertise. Hence, the effectiveness of transformation strategies ultimately hinges on the skills of the teams who develop and implement them, not how much it spends on AI tooling.