Next-Gen Software Engineering:
Humans define. AI builds.
The spec stack for AI coding.
Specifications are the new source code. To build reliable software with AI, we need a new foundation: a database to store specs, a compiler to formalize them, and a Git workflow to integrate it all.
SubLang: The Spec Compiler
Free expression, reliable generation.
SubLang bridges the gap between vague human intent and strict software logic.
Fuzzy to Formal
Transforms fuzzy source specs or chats into semi-structured formal specs (the target specs).
Semantic Consistency
Constructs the SubLang model to keep semantics consistent across your formal specs.
Dual-Readable
Generates a formal spec format that humans can audit and LLMs can perfectly understand.
but only their own profile"
Specbase: The Spec Management System
Plain text, multidimensional semantics.
The single source of truth for your software design and implementation.
Formal Spec API
Provides standard interfaces to store and retrieve human-readable formal specs.
Retrieval for AI
Allows AI agents to query or update the underlying model, not just raw text.
Data Integrity
Maintains data integrity across all formal specs as the project scales.
Tigs: The Git-Based Spec-Driven Dev Toolchain
Minimal primitives, flexible workflows.
Integrate spec-driven development directly into your existing Git workflow.
Chat-to-Git
Stores and manages source specs and chats directly in Git notes.
Version Control for Specs
Manages formal specs in Git through standard CLIs.
Agent Integration
Triggers spec generation, compilation, and maintenance with AI agents.
Try WellTested Today
Experience spec-driven development in action with our autonomous API testing tool.
Learn More About WellTested →