GitHub
Menu

Factory Development Lifecycle (FDLC)Open framework · v0.1

Build the factory that builds your software.

The Factory Development Lifecycle (FDLC) is the process for designing, assembling, governing, operating, measuring, and continuously improving the AI software factory that produces software.

SDLC

Governs the lifecycle of software.

FDLC

Governs the lifecycle of the factory that builds it.

Verified delivery, not merely generated code.

Understand FDLC in 5 minutesExplore the FrameworkMission Control →

FDLC / Category promise

Verified delivery, not merely generated code.Autonomy without evidence is just faster guessing.

Inside the factory

One autonomous software-delivery lifecycle.

An operating factory receives intent, coordinates bounded work, earns release through evidence, observes the result, and learns. These runtime steps are distinct from the lifecycle used to engineer the factory itself.

Interactive factory explorer

Follow one unit of work

Choose a stage. Then inspect the capabilities it activates.Swipe the lifecycle

05 / 09

Stage 05 / Evidence · Verification

Does the exact candidate satisfy the specification?

Evaluate the produced candidate independently and bind every finding to the exact subject.

State earnedCandidate verified
Durable record
Evidence · Verification
Evidence required
Current, attributable evidence plus independent checks against each acceptance and policy criterion.
Factory coordination
Collect evidence, run independent checks, preserve lineage, and fail closed on missing proof.

Factory capabilities

9 of 10 participate at Verify
Active capabilities are marked in green
Participates at Verify

Evals

Test the exact candidate against acceptance, quality, security, and policy criteria.

Explore the Trust area

This inner lifecycle produces software. The seven-stage FDLC lifecycle below evolves the factory that performs it.

The category boundary

Three categories. One critical distinction.

FDLC does not replace the SDLC. It engineers the persistent production system that increasingly performs parts of it. FDLC is the lifecycle for the factory itself.

CategoryTraditional SDLCAI-Native SDLCFDLC
Primary objectSoftwareSoftware + AI-assisted workflowThe software factory
Question answeredHow do we build software?How does AI change software development?How do we engineer the factory that performs autonomous software delivery?
01SDLC AI-Native SDLC

Agents move beyond code generation and participate across planning, design, implementation, testing, deployment, and maintenance.

02AI-Native SDLC AI Software Factory

Individual agent workflows become persistent, orchestrated production systems that transform intent into verified software with decreasing human intervention.

FDLC is the engineering discipline for that factory.

Architecture

Six areas. One governed system.

Models, agents, and tools can change. The factory keeps intent, identity, authority, state, evidence, decisions, and outcomes durable across every substitution.

Intelligence may vary.Authority remains explicit.Evidence must return.
01IntentDefine what should happen and how success will be judged.02HarnessMake execution durable, bounded, observable, and recoverable.03CapabilityPackage reusable ways for agents to act in a domain.04ModelMatch reasoning capability to the work without provider lock-in.05TrustDecide what may run, what can advance, and what requires judgment.06LearningImprove the factory from validated outcomes under change control.
Explore the reference architecture

Factory lifecycle / Outer loop

The factory is engineered, not installed.

FDLC evolves the production system from a measured opportunity to an operating factory line and then improves it from validated outcomes. Governance, security, observability, and measurement remain continuous throughout.

  1. 01Discover
  2. 02Design
  3. 03Assemble
  4. 04Validate
  5. 05Deploy
  6. 06Operate
  7. 07Improve
Continuous controls
Govern
Secure
Observe
Measure

Artifact protocol

Every transition earns its state.

The runtime lifecycle is made inspectable by durable nouns: intent, specification, authorized work, attributable attempts, evidence, decisions, releases, outcomes, and learning signals.

  1. 01Intent
  2. 02Specification
  3. 03Plan
  4. 04Work Order
  5. 05Attempt
  6. 06Evidence
  7. 07Verification
  8. 08Approval
  9. 09Release
  10. 10Outcome
  11. 11Learning Signal
01

The Plan is governed.

Approval binds one exact, reviewable revision.

02

Evidence gates transitions.

Missing, stale, or unrelated proof cannot become success.

03

Verification is separable.

A producer should not be its only verifier.

FDLC Manifesto

“Human attention belongs at consequential boundaries.”
Read all ten principles
  1. 01Intent must be explicit before execution begins.
  2. 02The Plan is a governed artifact.
  3. 03Every agent has an identity, sponsor, and bounded authority.
  4. 04Execution must be isolated, observable, and recoverable.
  5. 05Producers should not be their only verifier.

Open-source reference implementation

Mission Control

A governed control plane for planning, executing, verifying, approving, and learning from autonomous software work.

  1. 1Intent becomes an approved Plan
  2. 2Plans release bounded WorkOrders
  3. 3Attempts run under explicit authority
  4. 4Independent evidence gates acceptance
  5. 5Outcomes become governed learning signals
Learn moreGitHub ↗
MISSION CONTROL / COMMAND CENTERDEMO VIEW
Mission Control command center showing ranked work that needs operator attention
Deterministic demo view. Capability status is documented separately.

Factory maturity

How mature is your software factory?

Maturity is not the number of agents you run. It is the strength of your intent, execution, evidence, governance, and learning systems.

  1. L0
    Ad Hoc PromptingPersonal experimentation
  2. L1
    Assisted DevelopmentHuman-led work with AI support
  3. L2
    Governed AgentsDelegated units with explicit controls
  4. L3
    Orchestrated WorkflowsDurable governed workflows
  5. L4
    Verified Factory LinesOutcome-producing factory lines
  6. L5
    Continuously Learning FactoryGoverned continuous improvement
Assess your factory

The field guide

Learn the discipline from first principles to production.

The full 44-chapter AI Software Factory Guide covers architecture, governance, agents, harnesses, tools, context, models, verification, security, operations, and governed learning.

Choose a reading path

The FDLC.ai ecosystem

One discipline. Five distinct roles.

The open foundation defines the category. Products and services make it concrete without becoming the definition of FDLC itself.

01 / Open framework

FDLC

Defines it.

The open operating model, engineering discipline, and reference architecture.

Explore →02 / Available reference

Mission Control

Demonstrates it.

The open-source reference implementation with current capability boundaries visible.

Explore →03 / Open methodology

The Guide

Teaches it.

The methodology and engineering practices from first principles through operation.

Explore →04 / Design partner

FDLC Deploy

Implements it.

A forward-deployed path for proving one bounded factory line in a real environment.

Explore →05 / Proposed

FDLC Enterprise

Industrializes it.

The proposed commercial control plane for organizational scale and reliability.

Explore →