Orchestrated Agentic Programming

A governed path from AI capability to reviewable software delivery

OAP helps small teams use high-autonomy coding agents without giving up human control of architecture, risk, evidence, acceptance, or release. The method has been developed through practical SLAIF software projects and taught through highly demanded SLAIF workshops.

The documentation is organised in two parts: the main OAP operating discipline and the advanced Concentrated OAP mode for deadline-driven, high-intensity execution.

Read Chapter 1: Orchestrated Agentic Programming Read Chapter 2: Concentrated OAP

Overview

A human-governed operating discipline for delegating substantial software-development work to AI agents while preserving human authority over product intent, architecture, risk, evidence, acceptance, and release. OAP combines strategic planning, bounded execution, repository-centred evidence, independent review, and explicit release gates.

Outcome

SMEs and development teams gain a repeatable way to use high-autonomy coding agents without losing control of scope or software quality. Work remains reviewable through repository changes, meaningful tests, documented evidence, closure audits, and explicit human decisions.

Service Card

Category	AI development and deployment
Domain	Agentic Software Engineering, Software Delivery Governance, AI-Assisted Development
Vertical	Cross-domain
Target users	SMEs and Start-ups, Software Engineering Teams, Technical Leads, R&D and Innovation Teams, AI Training Providers
Service type	Methodology / Governed Development Workflow / Training Programme
Outputs	Repository governance instructions; bounded work orders; implementation changes; test and verification results; evidence packages; review records; closure audits; release recommendations; training materials.
Requirements	A version-controlled software project, development and test environment, access to a strategic AI and coding agent, and a competent human owner with authority over architecture, risk, acceptance, and release.
Limitations	OAP does not guarantee correctness, security, regulatory compliance, or production readiness. Evidence quality depends on acceptance criteria, tests, review discipline, and runtime containment. Human release authority remains mandatory.
Availability	Published methodology and training materials available; SLAIF workshops and controlled adoption support ongoing
Licensing	Method and training-material reuse terms to be confirmed; third-party models, agents, platforms, and dependencies retain their respective licences.