Alpaclaw connects to residents and operators through multiple messaging platforms simultaneously.
Channel
Status
Configuration
WhatsApp
Active
Allowlist mode, number: 424-234-1750
Telegram
Pending
Pairing DM policy — awaiting bot token from @BotFather
Discord
Pending
Will share PAI bot token initially, split later
Slack
Unconfigured
Available but not yet connected
Google Chat
Unconfigured
Available but not yet connected
4 Multi-Agent Hierarchy
Alpaclaw delegates complex tasks to specialized sub-agents, each with their own model and scope.
Alpaclaw (CEO / Orchestrator)
│ Model: Gemini 2.5 Flash
│ Role: Property management, resident comms, smart home
│
├── Research Sub-Agent
│ Model: Gemini 2.5 Pro
│ Role: Deep analysis, market research, complex reasoning
│ Scope: Read-only, no external actions
│
├── Automation Sub-Agent
│ Model: Gemini 2.5 Flash Lite
│ Role: Batch processing, status checks, monitoring
│ Scope: Pre-approved automation patterns only
│
└── [Future] Smart Home Agent
Model: Flash
Role: Direct IoT device control
Scope: Sonos, Govee, Nest, Tesla commands
5 Model Routing & Cost Optimization
Tasks are routed to the cheapest capable model, saving 80%+ on token costs.
Task Type
Model
Cost (per 1M tokens)
Orchestration (routing, delegation)
Gemini 2.5 Flash
$0.15 / $3.50
Simple tasks (status, formatting)
Gemini 2.5 Flash Lite
Cheapest tier
Complex reasoning (analysis, strategy)
Gemini 2.5 Pro
$1.25 / $10.00
Image generation
Gemini 2.5 Flash Image
~$0.039/image
Cost Strategy: Route 80%+ of requests through Flash/Lite. Reserve Pro for explicit complex reasoning. Cache data in workspace memory files. Batch API calls. Use one-topic sessions to avoid context bloat.