Language

Group Sites

Product AI-KIT

v-AI Kit — raising development quality with AI and a standard process

What it does Design documents and source code stay 100% in sync, automatically

Built on the principle of documents first, code second, v-AI Kit is a toolkit that codifies commands, agent skills, templates, and development rules. It plugs straight into the CLI agent in your terminal or IDE.

Watch the demo

$ /research new feature: upload PDFs to the forum
→ Gathered three or more pieces of evidence and fixed the scope of the research
$ /design BD → /design DD
→ SRS, basic design, and detailed design generated → awaiting review
$ /execute u0026u0026 /validate
✓ Zero Critical Issues

Memory bank syncing Docs ⇄ code 100%

0 % Less total effort, at best A 30–50% reduction across design, implementation, and testing
0 % Shorter delivery time, at best Twice the speed to market
0 % Design docs and code stay in sync automatically Every change flows back into the design documents

Product Overview

v-AI Kit in 85 seconds

v-AI Kit is a toolkit that puts AI to work as an assistant and keeps development on a predefined process. It codifies commands, agent skills, templates, and development rules, and integrates directly with the CLI agent in your terminal or IDE.

The AI works to defined processes, templates, and standards; each step’s output is reviewed and approved by the project team before the next step begins.

Documents first, code second
The memory bank preserves context
100% automatic sync between design docs and code
Zero Critical Issues

Scene

Demo video: from requirements input through automatic design-document generation, implementation, validation, and a zero-critical-issue completion report.

Business Impact

AI-driven results that change the business

Optimise cost and get your product to market twice as fast.

up to 0 %

Less total effort

Effort shrinks across every phase, from requirements analysis through validation.

up to 0 %

Shorter delivery time

Fewer reviews and less rework mean twice the speed to market.

Effort compared, phase by phase

Conventional With AI

Design -30% Conventional 100 With AI 70 SRS, BD, and DD generated automatically
Coding -50% Conventional 100 With AI 50 Code generated to your standards
Testing -40% Conventional 100 With AI 60 Six AI specialists validate in parallel

Indexed against conventional effort at 100. Actual figures vary by project size and technical domain.

Challenge u0026 Solution

Real-world development: the problems, and the answer

Simply having AI write code does not raise quality. If anything, it widens the gap from the specification and piles up technical debt.

Problem

How AI coding is used today

The cost of code-first, documents-later — or no documents at all
AI-generated code lacks consistency — every run produces something different
Code that merely runs, with no quality control behind it

Technical debt

Maintenance becomes progressively harder
Technical debt accumulates fast
Drift from the requirements, and enormous effort sunk into reviews and bug fixes

Solution

v-AI Kit・AI-Assisted Development

Treat AI as an assistant and develop against a defined process

Through its memory bank, v-AI Kit retains the context of the entire engagement and holds documents and source code rigorously consistent with each other. Each step’s output passes review and approval before the next one starts.

Memory bank Project context persists across sessions, so the AI always works from the same assumptions

Documents first, code second
Commands, skills, templates, and development rules, all codified
Every step passes review and approval before the next
Documents and source code stay consistent

Inside the Kit

v-AI Kit puts in your project

Commands, agents, skills, design templates, coding standards, and the memory bank. All of it lives inside the repository, so the whole team develops against the same standards and the same context.

Documents first, code second
Integrates directly with Claude Code
The whole team develops to the same standards

The AI works to predefined processes, templates, and standards. Each step’s output is reviewed and approved before the next step begins.

project
.claude/
commands/ Slash commands (/design, /plan …)
agents/ Specialist agents (backend, security …)
skills/ Skills that support the workflow
docs/
design-standards/ Design document templates (SRS, BD, DD)
coding-standards/ Coding standards per framework
memory-bank/ Context preserved across sessions
docs/
architecture/ Architecture docs generated by /architecture
design/ Design documents generated by /design
.vconf/ Configuration generated from config templates

How it works

Six commands that drive the development process

The AI executes; the developer decides. Each step’s output passes review and approval before moving on.

Phase 1: analysis and design

Pin down what to build, backed by evidence and design documents

/research

Research

AI The AI gathers information, investigates the requirements, and presents at least three pieces of supporting evidence. Developer The developer verifies that the data sources hold up and fixes the scope of the research.

/innovate

Innovate

AI The AI proposes three or more architecture options and analyses the trade-offs of each. Developer The developer weighs feasibility, then selects and commits to the best option.

/design

Design

AI The AI automatically generates the standard design documents — SRS, basic design (BD), and detailed design (DD). Developer The developer reviews them and confirms and approves the design before any coding starts.

/plan

Plan

AI The AI builds a detailed execution plan from the approved design. Developer The developer reviews it and settles the execution process step by step.

Phase 2: coding and testing

Build to the approved design, then validate it

/execute

Implementation

AI The AI generates code from the project plan and the coding standards. Developer The developer reviews the code and corrects anything that doesn’t hold up.

/validate

Validation

AI Six AI specialists validate in parallel across security, performance, and logic. Developer The developer confirms the Zero Critical Issues report and finalises the deliverable.

Developer (decides) AI (executes) Process (predefined) High-quality deliverables

System Architecture