AI-native · the analytical base as source of truth

The analytical basea new paradigm forbuilding software

Requirements, use-cases and the domain model as a structured semantic base in git. From it, Anna generates documentation, visual previews, cheap prototypes for fast value validation and production code — no rewriting, with easy iteration on every change.

I'm interested How it works

Requirements · Use-cases · Domain model · Tests · Code — one consistent whole

How it works

One source of truth, any output

The analyst models in a semantic base. Anna validates, indexes the dependency graph and generates everything else from the base — no manual rewriting.

Anna in action

anna — analytical base

 $ anna validate  
 ✓ 217 artifacts · 0 errors · 0 warnings  
 $ anna coverage --gap-type uc-without-tc  
 ⚠ 3 use-cases without a test-case  
 $ anna gap-fill --gap-type uc-without-tc --apply  
 ✓ 3 TC suggested → written to base  
 $ anna trace fr-uc-ent --format html  
 ✓ traceability matrix → html-preview/ ▋ 

.md

Analytical base

.md files · git-versioned · graph.json

Anna syncs and generates from the base:

anna-core engine

Engine — parser, validator, indexer, AI analysis

anna-ea sync

Bi-directional Enterprise Architect sync

anna-html preview

Static site with interactive diagrams

anna-confluence export

Automatic export to Confluence

anna-code code

Prototyping and production code generation

Comparison

The classic approach vs. Anna

What changes once the analytical base is the source of truth.

Classic With Anna

Source of truth ✕Proprietary EA format ✓Plain-text Markdown base, git-versioned

Documentation ✕Goes stale the moment it is written ✓Living docs — auto-generated

Consistency ✕Depends on analyst discipline ✓Validator + AI agent actively guard it

Cross-reference ✕Only in the analyst’s head ✓Explicit, indexed graph

Change impact ✕Unknown — risky on every change ✓Change-impact analysis over the graph

Prototypes ✕Hand-drawn outside the analysis ✓Clickable prototype straight from the base

Code from analysis ✕Manual translation into code ✓Incrementally generated backend and frontend

AI-native

Built for an AI agent from the ground up

Every artifact is machine-readable, cross-references are explicit and indexed, domain knowledge is bundled. Claude Code works directly with the base.

Semantic search

"find every requirement about authentication"

Automatic validation

Deterministic rules + LLM review keep the whole base consistent

Artifact generation

From a requirement it drafts a use-case, from a use-case a test-case

Code generation

From entities: JPA entities, services, controllers and integration tests

Impact analysis

"what breaks if I change this entity?"

Incremental builds

Recomputes only changed artifacts via the dependency graph

Like what you see?

Anna isn't publicly distributed yet. If it caught your eye, get in touch — I'll gladly show more and discuss how it could help your team.

No spam. I'll only contact you about Anna.