Skip to content
Delivery process

A repeatable delivery process for enterprise AI agents and AI applications.

Run a four-stage cycle that keeps agent design, supervision, evaluation evidence, and release readiness aligned. Delivery breaks down when planning, execution, evidence, and release readiness are split across separate products and teams.

Browse the full user guideRequest a demo
Delivery stages

Four stages, each with clear activities and evidence.

Define activities, outcomes, and delivery evidence that travel with the work.

Stage 1
Planning (define)
In this stage, the foundation is set by defining the AI application's purpose, scope, and success criteria.

Key activities

  • Define Business Vision and Success Metrics
  • Stakeholder Onboarding and Kickoff
  • Identify User Personas and Scenarios
  • Define Solution Requirements and Constraints

Artifacts & evidence

17 artifact templates

User guide

Stage 2
Exploration (shape)
In the Exploration stage, the team shapes how 'good' can be achieved by translating plans into actionable desi…

Key activities

  • Develop User Stories and Draft Flows
  • Define Data Requirements and Sources
  • Evaluate and Select Tools/Integrations

Artifacts & evidence

11 artifact templates

User guide

Stage 3
Development (build)
In the Development stage, the team builds and expresses 'good' in a working AI application.

Key activities

  • Define Agent Persona, Purpose, and Scope
  • Architect Agent Components and Integrations
  • Develop and Refine Prompts and Instructions
  • Test and Refine Agent Behavior
  • Design Conversation UI/UX
  • Test End-to-End Conversation Experience

Artifacts & evidence

23 artifact templates

User guide

Stage 4
Evaluation (prove)
The Evaluation stage serves as the formal proof of 'good' – assessing the AI application against the success c…

Key activities

  • Define Evaluation Criteria and Test Plan
  • Run Functional and Scenario-Based Evaluations
  • Analyze System and Model Metrics
  • Conduct Governance and Safety Review
  • Approve for Production and Monitor Rollout

Artifacts & evidence

16 artifact templates

User guide

Process module

The delivery playbook, operationalized inside Gaia.

The Delivery Management module turns the framework into cycles, stage tracking, and evidence capture so teams stay aligned from kickoff through evaluation and release.

Module
Delivery cycles
Launch a cycle from the four-stage playbook and keep each project aligned.
  • Create cycles in minutes
  • Track active stage & status
  • Archive cycles for history

User guide

Module
Stage playbooks
Initialize recommended activities and keep each stage auditable and on track.
  • Playbook-driven activities
  • Stage status and reopen flow
  • Role assignments at a glance

User guide

Module
Evidence & artifacts
Capture structured evidence using templates tied to every activity.
  • Template-driven evidence
  • Markdown notes & attachments
  • Export-ready documentation

User guide

Module
Delivery tasks
Connect delivery cycles to task boards so owners can execute with accountability and risk visibility.
  • Assign tasks to owners
  • Track delivery status and risk actions
  • Keep delivery work accountable

User guide

Bring the delivery playbook into every project.

Use the process module to keep roles, activities, and evidence aligned from kickoff through ModelOps-style evaluation and release decisions.

Explore platform modulesRequest a demo