Project Overview

Agentic Software Factory

Hitting process ceilings with Claude Code? The Agentic Software Factory makes AI-assisted development more structured, reproducible and team-ready. Your project idea turns into running software – with a clean Git history, supervised runs and traceable approvals.

local control plane

markdown artifacts

run phases

agent roles

🚀 Request Early Access See licensing

Who is this for?

Built for technical stakeholders who want to adopt AI-assisted development in a structured way. Not a mass-market consumer product but a productive working surface.

Primary

Software architects

Clear guardrails, traceable decisions, Git and build discipline as technical gates.

Primary

Lead developers

Efficient Claude usage without prompt chaos. Reusable project setups instead of repeated scaffolding.

Primary

Tech leads

Transparency over runs, logs and approvals. Sign-offs are deliberate, not rubber-stamped.

Good fit

Small product teams

Quickly usable artifacts instead of dead documents. Follow-up runs instead of overly long single runs.

Coming soon

Enterprise teams

Auditability, multi-tenant operating models and compliance scenarios are on the roadmap for later versions.

Which problems does it actually solve?

Using Claude Code directly in the shell is powerful but procedurally rough. The platform addresses exactly these friction points:

Prompt chaos in the shellProject definitions emerge ad hoc, are not versioned, and are hard to reuse.
Few commits, muddy Git historyLong Claude sessions produce large, hard-to-review changes.
Approvals without contextManual approvals get rubber-stamped; critical phases are not visibly flagged.
No end-to-end traceabilityLogs, build results and agent decisions scatter across terminal, editor and notes.
Follow-up projects start from scratchProven agent roles, teams and guardrails are not preserved.

A look inside the platform

Schematic preview of the main working areas. The final visuals may differ – the structures are in place.

Project wizard

Structured intake instead of an empty markdown file.

Run view

Phases, status and current step at a glance.

Git & build

Commits, working tree and build status right in the UI.

Markdown artifacts

Generated specification files as an editable preview.

What it looks like in practice

Say you want to build a simple REST API for invoices. Here is the path through the Software Factory – from a 15-second wizard to a validated commit.

Your input · 15 seconds in the wizard

Project name: Invoice API
Target:       REST API for invoices,
               CRUD + state transitions
Tech:         Spring Boot 3, PostgreSQL, Flyway
DoD:          all endpoints tested,
               mvn verify green, coverage ≥ 80%

What the platform produces

✓ PROJECT.md            47 lines
✓ INSTRUCTIONS.md       32 lines
✓ WORKFLOW.md           19 lines
✓ DEFINITION_OF_DONE.md 14 lines
✓ AGENTS.md             21 lines
✓ README.md             28 lines
✓ Git repo              initial commit
✓ Run template          mvn verify gate

→

Then click Start run. What happens next is fully visible in the UI, live:

Run #42 · Invoice API · claude-code adapterEXECUTION

14:32:01✓INTAKE2 s

14:32:03✓PROMPT_ASSEMBLY4 s

14:32:07✓WORKSPACE_PREPARATION8 s

14:32:15⟳EXECUTION (claude-code)4 min 12 s

├ src/main/java/… +247 lines

├ src/test/java/… +89 lines

└ 3 commits · feat, test, docs

14:36:27✓VALIDATION mvn verify52 s

14:37:19⏸COMPLETION (awaiting approval)—

Total: 5 min 18 s336 lines of code42 testscoverage 82%

What the project contains

Project Definition

Title, target picture, constraints, tech preferences and DoD.

Agent Definition

Roles, default instructions, tool profiles and limits.

Team Assembly

Agent teams for different project types and run flavours.

Prompt Assembly

Automatic generation of PROJECT.md & co.

Run Orchestration

Start, monitor, pause, complete.

Git & build

Commits, working tree and build results in the UI.

How the user works

1. Capture the ideaStructured fields plus free text.

2. Generate artifactsMarkdown files like PROJECT.md.

3. Prepare workspaceDirectory, repo and run context.

4. Start the runClaude Code as the dev agent in the backend.

5. Monitor the runPhases, logs, commits, builds.

6. InterveneApprove, correct, start follow-up run.

Licensing · at a glance

Three tiers with a deliberately low-friction entry point. Detailed comparison and registration flow on the Licensing page.

Criterion	Demo	Community	Full
Registration	no	free	admin-only
Mock adapter	✓	✓	✓
Claude adapter	–	✓	✓
Runs per month	5	20	unlimited
Minutes per run	5	10	unlimited
Agents per team	1	3	unlimited
Offline usage	always	up to 7 days	up to 7 days
Status	available	available	coming soon

Technical foundation of version 1

Java 21Spring BootThymeleafHTMXPostgreSQLSpring SecurityKeycloakDocker ComposeJUnit 5

Angular is deliberately not required for version 1. Focus on workflow, orchestration, traceability and reusability.

What V1 deliberately does not yet deliver

parallel multi-agent execution
a graphical workflow designer
cloud scaling and multi-tenancy
a full CI/CD platform

Recommended domain model

ProjectDefinitionInstructionSetDefinitionOfDoneAgentDefinitionAgentTeamRunRunPhaseTaskExecutionStepExecutionLogApprovalPolicyApprovalDecisionGitRepositoryGitCheckpointBuildResultArtifactPromptArtifact

Key UX design decision

The user should not have to hand-write markdown files first.

Idea → structured UI capture → auto-generated markdown artifacts → editable preview → run

Why this matters: This is precisely what makes the platform simpler and more productive than using Claude Code directly in the shell.

Architecture & background

Deep Dive: architecture behind the Software Factory

The Software Factory is built on a thought-through architecture for agentic software development. The companion whitepaper describes reference architectures, agent orchestration, guardrails and SDLC integration across 76 pages – as the conceptual foundation of the product.

Open whitepaper page Open PDF directly

Be one of the first users.

The Software Factory is under active development. If you’re hitting limits with today’s AI coding tools and want a structured alternative: drop me a line – I’ll walk you through the system personally and reply within 1–2 business days.

🚀 Request Early Access Email me directly