- Choose this when
- The system is custom code you own, it still carries the business, and it is now the thing slowing you down.
- This is not
- Upgrading a content or commerce platform you bought. That is Digital Experience and Commerce.
- Platforms
- Evidence
- Ask about the work behind this capability.
AI can only act on systems it can read.
A system is readable when your engineers, and the AI working with them, understand it well enough to change it safely. That takes three things: current knowledge of the code, test coverage to catch the mistake, and a release cycle short enough to learn from. A system that has carried a business for years has often lost at least one of these, and often all three. None of that is a model problem, and none of it is fixed by buying a better one.
The model is rarely the constraint.
If your pilots worked and nothing reached production, it is often one of four things. The application cannot be changed safely. The data cannot be trusted. The release cycle is too slow to matter. Or the platform the work sits on is closed. We remove the constraint first, then put AI to work on the other side of it. AI applied before the constraint is removed is how an AI budget disappears.
Build
First, make the system readable, and keep it that way.
Three capabilities that remove the constraint and keep it removed. We keep the tests and the knowledge of the code with your account.
- Choose this when
- The thing does not exist yet, and it will be your code when it does.
- This is not
- Configuring a platform you bought. That is Salesforce, or Digital Experience and Commerce.
- Platforms
- Evidence
- Ask about the work behind this capability.
- Choose this when
- Releases are slow, bugs are escaping, or the cost of testing has become the reason not to release.
- This is not
- Security testing, which is Security and Compliance Engineering. Running quality as an ongoing service, which is Managed Services.
- Platforms
- Evidence
- Ask about the work behind this capability.
Apply AI
Then put AI where the work is.
Custom builds on your data, the data underneath them, and the platforms you already pay for.
- Choose this when
- The answer has to come from your systems and your documents, not from a general model.
- This is not
- AI features inside a platform you already own. That is Intelligent Automation.
- Platforms
- Evidence
- Ask about the work behind this capability.
- Choose this when
- The work is on the data itself: the pipelines, the warehouse, the reports, the models.
- This is not
- Setting up a marketing analytics tool. That is Digital Experience and Commerce.
- Platforms
- Evidence
- Ask about the work behind this capability.
- Choose this when
- The work runs on robotic process automation, business process management, or Microsoft Copilot.
- This is not
- Agents built on your own data. That is AI Agents and Assistants.
- Platforms
- Evidence
- Ask about the work behind this capability.
Platforms
Fix how it runs and ships.
Where the work sits, how it gets there, and how it connects.
- Choose this when
- The job is getting onto the cloud, or re-engineering what is already there.
- This is not
- How software ships once it is there, which is Platform Engineering. Securing it, which is Security and Compliance Engineering.
- Platforms
- Evidence
- Ask about the work behind this capability.
- Choose this when
- The delay is between a change being written and a change being live.
- This is not
- Moving to the cloud in the first place. That is Cloud Migration.
- Platforms
- Evidence
- Ask about the work behind this capability.
- Choose this when
- Salesforce is the platform the work sits on.
- This is not
- A custom application built outside Salesforce. That is Product Engineering.
- Platforms
- Evidence
- Ask about the work behind this capability.
- Choose this when
- The platform is a content, commerce, or marketing analytics product you have bought.
- This is not
- Custom-built web applications. That is Product Engineering.
- Platforms
- Evidence
- Ask about the work behind this capability.