07 — ROADMAPA credible twelve months

The phased plan.

From team adoption to operating system, in five phases. No big bang. No expensive bets before evidence. Each phase delivers value while building the foundation for the next.

PHASE 01 · Months 1–2
Claude adoption & weekly demos
$2k/mo

Goal · Get every PBVR leader using a frontier AI tool daily. Build muscle memory before building systems.

Build items
  • 01Anthropic Claude Teams licenses for ops, marketing, finance
  • 02Weekly 30-min internal AI demo from a different team member
  • 03Shared prompt library in Notion
  • 04First MCP server: read-only access to the property docs folder
Business value

Habit formation. Surfaces real use cases. Trains intuition for what AI is and isn't good at — before any vendor pitch.

Risks to manage
  • !Adoption uneven across roles
  • !Prompt copy-paste fatigue without a library
Leadership decisions
  • Claude Teams vs Claude Enterprise tier
  • Who runs the demo cadence
  • Acceptable use policy (one page)
PHASE 02 · Months 3–4
Connect top sources of truth
$4k/mo + $18k impl

Goal · Wire MCP servers to our highest-leverage data: PMS, accounting, drive, calendar. Read-only first.

PHASE 03 · Months 5–6
Retrieval layer over key documents
$5k/mo + $22k impl

Goal · Vector index our 29 years of property files, past guest messages (consented), maintenance logs, brand voice docs.

PHASE 04 · Months 7–9
First live workflow: guest-message triage
$7k/mo + $35k impl

Goal · Ship the first real AI in front of a real workflow. Guest message comes in → AI drafts → human reviews → sent. Logged.

PHASE 05 · Months 10–12
Maintenance dispatch & owner reporting
$10k/mo + $45k impl

Goal · Extend the same pattern to two more workflows. Build operating leverage beyond the inbox.

Total year-one spend
~$420k
Software + impl, all phases · directional
First measurable ROI
Month 9
Live guest-message triage workflow
Decisions deferred to year two
5
Dynamic pricing, voice, local guide, others
Now: the explicit list of what to do this quarter.