Skip to content

From DTAP to BDTAP: software starts with the business

The classic DTAP pipeline starts with Development. With AI, software increasingly starts before IT, with the business itself. That is why we add a B: Business, Development, Test, Acceptance, Production. In Dutch we call it BOTAP.

DTAP is old-school. Now it is B(usiness)DTAP.

DTAP assumes that software starts with IT: the business supplies requirements, IT spends months turning them into software. But with AI, someone who truly understands the business problem can express the idea in working software: a prototype, a small application, an experiment. This is already happening at scale. Gartner expected that by 2026 at least 80% of users of low-code tools would work outside the formal IT department, and that by 2027 75% of employees would acquire, modify or create technology outside IT's visibility.

B → D → T → A → P

  • B, Business: the business builds itself, with AI, prototypes and small applications. IT provides the means (access to data, test data or synthetic data, and the tools), but does not build the application. The goal is not to turn everyone into a developer, but to make the idea visible in working form.
  • D, Development and operationalisation: only when an application becomes important and is approved to be taken into management does the code move to Development. From there IT handles application management, as in the familiar DTAP pipeline.
  • T, A, P: the familiar steps of test, acceptance and production.

The lesson of Access

In the 1990s Microsoft Access made the same promise: you do not need to be a programmer to build an application. That worked wonderfully for speed and innovation, until such an application became business-critical and nobody knew who maintained it, how it was secured or how to scale it. AI threatens to cause exactly the same problem, on a far larger scale. Hence our key sentence: AI democratises software design, not software management.

IT does not disappear, its role shifts

Instead of spending months translating requirements, IT sets up the safe playing field in advance: approved data sources, APIs and MCP servers, authentication and authorisation, architecture rules and permitted tools. Data belongs to that too: access to real data where allowed, and otherwise test data or synthetic data. That way the business can work with data that is permitted for the trial. Within that space the business can experiment freely. That is how you prevent shadow AI without slowing innovation down.

From B to D: the most important step

As soon as something proves its value, it moves from B to D. IT then does not ask "how would we have built this ourselves?", but: "the business has proven this works; how do we make it fit for the organisation?" Architecture, security, dependencies, performance, cost, ownership, logging, documentation and management are checked and, where needed, refactored. AI helps there too: not only to build something quickly, but to analyse a business solution and hold it against the enterprise architecture.

Technology is a life cycle, not a religion

For the step from B to D, the technology matters a great deal. If the development pipeline is built on your own software, ordinary open code, taking it into management is simple: IT takes over the same code, improves and extends it, and fits it into the technical IT landscape. If a solution has to move to OutSystems or Mendix, that means a rebuild in the platform. The prototype then serves as a working specification, but the code does not come along. Choose that only if the platform's security and governance are worth the rebuild. If the functionality later becomes a commodity, it can disappear into SAP, Microsoft, Salesforce or another standard solution.

Three key sentences

  • Logic first, then AI. What ordinary code can calculate, you never leave to a model: deterministic first.
  • AI democratises software design, not software management.
  • The business owns the problem, IT owns the guardrails, the organisation owns the life cycle.

How we help

Two Swallows helps you set up BDTAP:

  • The playing field for the business: approved data sources, APIs and MCP servers, authentication and authorisation, permitted tools and house rules.
  • Data for the business: access to the right data sources, and test data or synthetic data where real data is not allowed.
  • A development pipeline on your own software, so that the step into management does not require a rebuild.
  • The step from B to D: agreements on when a solution is valuable enough, and who then becomes its owner.
  • The D phase itself: analysing proven solutions and making them fit for production, with AI and our AI Architecture Guidelines.
  • The technology per phase: your own code, OutSystems or Mendix, or a standard solution.

What to watch out for

  • Set up the playing field before the business starts, not afterwards.
  • Agree in advance on the moment from B to D: when is something valuable enough?
  • Record ownership as soon as a solution is used by more people.

Would you like advice on your environments and your release process? Request a free advice conversation.

Sources

Updated on

Services

What is full stack?

Full stack means every layer of an application, from the screen to the database and the server. A full-stack developer or team works across those layers.

Services

Who holds up a mirror to your CIO?

CIOs and IT managers divide their time between budgets, suppliers, projects and people. A sparring partner with current technical knowledge makes their decisions stronger, and gives the board confidence.

Services

What is the Definition of Done?

The Definition of Done records in advance when something is really finished. That way everyone knows in advance what the delivered work will be checked against.