Breathe AI Launchpad or Dynamics 365 – when is building your own worth it?

We deliver both. We implement Dynamics 365 and Power Platform for customers every week, and we built Breathe AI Launchpad for the cases where the standard platform is not the answer. That makes us neither neutral nor dishonest – but it does mean we have seen both mistakes: organisations building their own when they should have bought standard, and organisations forcing the standard platform into something it was never meant for.

Here is how we tell them apart.

The key difference is not technology. It is how cost scales.

Dynamics 365 and Power Platform are priced per user, per month, forever. That works well when every user extracts value every day. It works badly when a lot of people need to touch an application occasionally.

Microsoft's published list prices make the point:

Note too that Microsoft removed the “per app” plan from the licensing guide in January 2026. The cheap entry point for a single application no longer exists in the same form – in practice, per-user licensing is now the main way in.

Breathe AI Launchpad is priced differently: a fixed monthly team fee for the development environment and governance, plus a flat fee per application we run in production. Cost therefore follows how many applications you run – not how many people use them.

If you have 40 staff who each need to record something in a line-of-business application a couple of times a month, the 41st user costs nothing under a per-application model. Under per-user licensing every one of them costs full price, every month, however little they use it.

The reverse also holds: if you want to run eighteen small applications for one small team, the maths flips. There, the standard platform is cheaper.

When Dynamics 365 and Power Platform are the right answer

This is the part suppliers with their own product tend to skip, so let us be explicit. Choose the standard platform when:

• You need a business process that already exists. Sales pipeline, case management, finance, field service. Do not rebuild CRM. Dynamics 365 has decades of process logic built in, and it is nearly always better than what you can specify in the time available.
• The need is 80% standard. If out-of-the-box functionality covers most of it, configuration is cheaper than building from scratch.
• Users are few and intensive. Twelve salespeople living in the system all day comfortably justify a per-user licence.
• You do not want to own an application. Microsoft ships new features, security updates and platform work without you lifting a finger. That has real value.
• The people building are not developers. Power Platform exists so a controller can build something useful without opening a repository.

When building your own pays off

• The need is genuinely distinctive. No standard product covers it, and you have already tried.
• Many occasional users. See the maths above.
• External users. Customers, suppliers or citizens who need access get expensive quickly under portal licensing models.
• You want to own the code. Launchpad applications are ordinary .NET applications in your own Git repository with your own CI/CD. The exit cost is a handover, not a migration.
• Data has to stay in your environment. Everything runs in the customer's own Azure tenant. No customer data passes through Breathe.

What AI-assisted development actually changes – and what it does not

It is easy to oversell here, so let us be precise.

AI coding agents have made writing code dramatically faster. A case-handling interface that used to be two weeks of work is now often a couple of days. That is a real shift, and it moves the build-or-buy line – things that were too expensive to build three years ago are not any more.

But what got fast is only one part. Understanding requirements, integrating with systems that will not cooperate, testing, security review, operations, fixing a fault at 11pm – none of that got meaningfully faster. The bottleneck moved from writing code to governing it.

That risk is exactly what Launchpad is built for: people build at full speed in a safe sandbox, and the business makes one decision – which prototypes go to production. What passes the gate, we industrialise and run.

Summarised in a table

The most common answer is “both”

In practice most organisations end up with a combination, and that is not a compromise – it is usually the right architecture. Dynamics 365 as the system for customers, sales and cases, where the standard process is good. Custom applications around it for what is genuinely specific to the business, and which would otherwise become an expensive customisation in a system that was not designed for it.

We have no interest in selling you a custom application if Dynamics 365 solves the problem better. That usually becomes clear within the first half hour of a conversation.

Microsoft list prices are taken from Microsoft's own pricing pages (Dynamics 365 Sales and Power Apps), August 2026. Prices in USD excluding VAT and subject to change.

David Kjell CEO & AI Solution Architect