Operational Framework
ASF
Augmented Software Factory
Mixed Human + AI Team · First delivery in 30 days.
ASF is B.I.T.'s operational framework for software development with AI agents. It blends human expertise with AI capabilities to accelerate delivery, maintain high quality, and ensure the client ultimately owns the factory without lock-in.
The SCRUM-AI™ Cycle
A compressed weekly cadence driven by AI speed and governed by explicit human decision gates.
Sprint Planning
Planning & Backlog
The PM and Dev Lead define Epics and sprint priorities. The Epic & Task Agent automatically generates Task.md (L3) files with granular acceptance criteria and prompt seeds. The Dev Lead validates each task before creating GitLab Issues.
Sprint Execution
Development & Continuous Testing
The Senior Developer works in local sessions with the Code Generation Agent (Claude Code) to produce code. The Test Generation Agent translates AC into Playwright/Pytest tests executed on every commit. The Code Review Agent (GitLab Duo) provides an initial MR review, finalized by the Dev Lead's signature.
Sprint Review
Demo & Staging Approval
The PM presents the working software on a staging environment to the client. The Documentation Agent automatically generates release notes and technical documentation directly from MRs. The client provides feedback or formal acceptance.
Retrospective
Pattern Promotion
The team analyzes sprint metrics and velocity. The Dev Lead identifies emerging best practices and decides whether to promote them into architecture.json, repository CLAUDE.md files, or at the company level, keeping the factory self-improving.
Change Request
Change Flow Management
Any requirement changes or critical bugs are analyzed by the RIB Delta Engine to estimate impact and effort. Upon PM and Dev Lead approval, a mandatory top-down update cascade is propagated across the Prompt Stack (product.md → RF-XX.md → EpicXXX.md → Task.md).
How ASF Works
An operational model built for speed, transparency, and progressive ownership.
Methodology
First Delivery in 30 Days
In the first 30 days, the first delivery is tested and deployed on staging. Methodology: SCRUM-AI: 1-week sprints. Prompt Stack L0→L3. Explicit human decision gates at every phase.
People & Roles
1-Week Sprints & Mixed Teams
Sprints last 1 week compared to traditional 2-week Scrum. People and roles: 4 human roles + 6 AI agents. Non-interchangeable perimeters: humans control the domain, agents provide speed and parallelism.
Platform
12-Month Engagement & Platform
In 12 months the engagement is complete, at the end the client owns the factory. Platform: GitLab Premium as Scrum infrastructure. CronStudio as UI layer. Claude API for code generation, testing, documentation.
Client Ownership
Uniform Cost & Zero Lock-in
125 euros/day/FTE uniform cost: human or AI agent. At 12 months the client owns the factory: documented methodology, trained developers, configured Prompt Stack, zero lock-in.
Accelerate your development.
Implement the Augmented Software Factory and start building with AI today.