AI Platform Roadmap

Stage 1: Architecture and Prototype Definition

Goal:

• lock the first real pilot workflow for TV Mount Australia
• define the review queue and holding-response logic
• define the first supported channel pair
• define the minimum viable shared core stack

Deliverables:

• architecture brief
• staged plan
• channel choice
• first prototype scope
• operator review design
• booking integration boundary

Stage 2: Internal Pilot

Goal:

• build a working internal prototype on business-scoped infrastructure

Recommended focus:

• TV Mount Australia first
• 2 to 3 internal pilot tenants as a target, but only one active critical workflow initially
• one primary virtual employee
• one review queue
• one workflow path
• one knowledge path
• one holding-response path

Stage 3: Limited External Pilot

Goal:

• prove value with a controlled real-world client or internal business use case

Recommended focus:

• controlled onboarding
• clear pricing assumptions
• stable operator review path
• Google Calendar booking workflow if stage 2 is stable
• feedback loop into prompts and workflows

Stage 4: Repeatability

Goal:

• make tenant setup and service rollout repeatable

Recommended focus:

• Ansible-based deployment and configuration
• tenant templates
• documented onboarding process
• clearer separation of shared vs tenant-specific components

Stage 5: Broader Productization

Goal:

• support multiple clients, channels, and agent roles safely

Recommended focus:

• stronger provisioning model
• stronger billing boundaries
• better support tooling
• future model-provider flexibility