Part III of our series on AI-assisted software development.
At the end of Part II of our series, we mentioned that the right answer is often both: custom-built and professional software used together. In this article, we explain what that can look like in practice.
The dividing line is not between companies, but between tasks
The question is rarely whether to use standard software or build a custom solution for the entire company. Instead, the question is: which part belongs where?
Everyone in our industry will recognise this principle from their day-to-day work. Hardly anyone builds their own truss system, but of course you might create a custom decorative structure for one particular customer. The same principle applies to software.
Internal processes—where employee data is stored, working hours are recorded and documentation is maintained—belong in a system that is continuously maintained, professionally operated and backed by someone who takes responsibility for it.
The external layer—everything built on top of this data—is well suited to custom development. This could be a specialised report, integration with your own CRM or a tool that represents a workflow unique to your company.
For this second layer to work, however, the first one must be open. And as a provider, we see it as our responsibility to ensure that it is.
Our API is open and always included
CrewBrain has a well-documented API that allows practically every type of object in the system to be read and modified.
Our customers demonstrate just how much this makes possible. Virtually any third-party application can be connected to CrewBrain through the API. Companies can also build their own workflows—for example, reconciling invoices and automatically paying freelancers. For many other tools, including rental management systems such as Eventworx, ready-made connectors are already available in CrewBrain.
The best part is that the custom-built tool does not need to store the sensitive data itself. It simply works with or updates the data.
This means there is far less need to worry about matters such as documentation requirements, backups and operational resilience because the underlying system remains in place as the foundation.
MCP: enabling AI to work directly with CrewBrain
For some time now, there has been a common standard that enables AI applications to communicate with other software. It is called MCP, and CrewBrain was one of the first providers in the industry to release an MCP server as an interface for AI systems.
In practical terms, this means that AI applications such as ChatGPT or Claude can work directly within CrewBrain. For example, they can create jobs for you, record absences and handle many other tasks. The usage figures for our MCP server show that some of our customers already use this option every day, and the feedback we have received so far has been excellent.
Two aspects are particularly important to us:
You decide which AI to use. Just as with our open API specification, we do not prescribe which models or specific software solutions may be connected to CrewBrain. We know that individuality matters in this industry and that data protection requirements vary considerably.
Companies that want to operate their own locally hosted AI can also connect it to CrewBrain. For businesses with strict internal policies or clients from the public sector, this is often a requirement.
You do not have to bypass CrewBrain to build your own solution. That is the central conclusion of this series. The desire to create a custom tool quickly is entirely understandable. But it does not need to result in a second data repository that nobody maintains or takes responsibility for when something goes wrong. With the API and MCP server, you can build on top of the system that is already running.
What we intend to do
For us, an open interface is only one important piece of the puzzle. There are three things we are taking away from this:
We want to see what you build. Your custom solutions reveal genuine requirements more quickly than any workshop or support email.
We want to remain honest about our roadmap. As much as we would like to implement every feature request, that is simply not possible. That is why our roadmap transparently shows which features we are not currently considering. These are often the best candidates for custom development using our API.
Your data belongs to you. CrewBrain remains open through its API, which is included in every plan without usage limits apart from generous rate limits. You can use it to access and modify your data at any time.
In conclusion
Two years ago, “build or buy” was an either-or question that was usually decided by the available budget. Today, the question is where exactly the dividing line should be drawn.
For us as a provider, this creates new challenges because openness and the ability to fit individual requirements suddenly matter just as much as the range of features. Nevertheless, we believe this development is positive for both sides.
And if you are currently building your own tool and are unsure how to connect it to CrewBrain, simply get in touch. Our support team is just as happy to answer questions about our API and MCP as it is to receive any other feedback.
1 comment
Comments are closed.