IDD concept: Specify intent
Intent Specification
An Intent Specification is a structured description of what should be built and why, complete enough that autonomous systems can execute against it while remaining aligned with human goals. It is the governing artefact of Intent-Driven Development, and it is built from six elements.
Why a specification, and why now
An agent working from an incomplete description fills the gaps with generic assumptions, and it does so quickly and without signalling where it has guessed. The Intent Specification exists to remove that guesswork. It captures, in human language, the information an autonomous system needs to build the right thing in the right way, and it gives everyone involved in delivery, from product and design through architecture, security and quality, a single shared point of coordination.
Teams will hold specifications in whatever tools suit them. What matters is not the format but that the specification covers each of the six elements, because each one closes a different gap through which misalignment enters.
The six elements
The elements grew out of the five questions that IDD originally asked before any implementation began. Integration with Domain-Driven Design made clear that domain context needed explicit treatment of its own, and the five questions became six elements.
-
Intent: the why and the what
Why is this being built, what problem does it solve, and what outcome defines success? Without a clear articulation of intent, everything downstream becomes interpretation.
-
Domain Context: the world
Which bounded contexts are involved, which entities, aggregates and value objects exist, and what ubiquitous language does the organisation use? When agents lack domain context they default to generic patterns, producing solutions that are technically correct and semantically wrong.
-
Success Criteria: the done
The functional and non-functional requirements, expressed where helpful as BDD acceptance criteria. Without explicit success criteria there is nothing to measure Intent Fidelity against.
-
Validation: the proof
What must be tested, which edge cases matter and how the system should behave when things go wrong. In IDD the tests are part of the specification rather than an afterthought, and when validation is well defined, agents can generate test suites that explore far more combinations than people typically consider.
-
Constraints: the boundaries
Technical platforms, architectural patterns, integrations, resource limitations and regulatory obligations. These are the operational realities within which any solution must work, and autonomous systems need them stated rather than implied.
-
Ethical Considerations: the should we
Should this system exist at all, who is affected by it and what harms might arise? When autonomous systems can build rapidly, the question of whether to build becomes as important as the question of how.
How much detail a specification needs
The level of detail should match the level of risk and complexity involved. A low-risk change to copy or styling needs very little; a payment flow or a system handling personal data needs a great deal. For teams beginning with IDD, writing down just the Intent and the Constraints already improves clarity considerably, and the remaining elements can be added as experience grows.
A deliberately comprehensive worked example, an abandoned cart recovery system for an e-commerce business, shows all six elements working together at the high-stakes end of the range. It includes domain model definitions, BDD acceptance criteria, validation scenarios and edge cases, performance benchmarks, compliance requirements and a structured ethical impact assessment.
Write shared intent once
Many constraints, such as security baselines, compliance obligations and architectural standards, apply to every project in a domain or across the whole organisation. Mature IDD organisations define these once and let individual specifications inherit them, so that a project specification references shared intent instead of restating it. The Intent Hierarchy describes how that inheritance works and why it is best introduced only once duplication becomes painful.
What a good specification enables
When the six elements are captured clearly, autonomous systems can execute reliably, cross-functional teams share an understanding of what is being built, alignment becomes measurable, governance becomes a matter of evidence rather than intuition, and institutional knowledge accumulates in a form that future work can reuse. None of this removes accountability from the people involved: delegating execution does not delegate responsibility.
Where it comes from
The six elements were set out in How Do You Specify Your Intent?, published on 17 March 2026, building on the five questions in the first IDD article. They form Chapter 4 of the ebook.