LlamaChilly
Core API
The LlamaChilly API is a sovereign, event-driven infrastructure designed to capture, qualify, and convert restaurant leads at scale. Built for European restaurant groups operating in high-volume environments.
- Base URL: https://api.llamachilly.com/v2
- All responses are JSON-encoded.
- Rate limit: 2,000 req/min on Pro tier, unlimited on Enterprise.
- Data residency: EU-West (Amsterdam). GDPR compliant by architecture.
Authentication
Bearer Token Security
Authenticate all requests with a Bearer token generated from your sovereign dashboard. Tokens are scoped, rotatable, and tied to specific network clusters.
Authorization: Bearer lc_key_v2_...- Tokens prefixed with lc_key_v2_ for v2 API access.
- Include X-Cluster header to route to your private cluster.
- Keys are never stored in plaintext. Hashed at rest with SHA-256.
- 2FA enforcement available for key generation.
The Lead Object
Core Entity
A Lead represents a qualified buyer signal captured by AI agents across channels. Each lead is pre-scored using multi-variate intent analysis and enriched with RAG context from your restaurant's knowledge base.
/v2/leads/{lead_id}- id: Unique identifier, prefixed with lead_.
- channel: Source of capture: whatsapp_pro, google_maps, instagram_dm.
- score: Integer (0–100) representing pre-qualification confidence.
- status: One of: RAW_SIGNAL, PRE_QUALIFIED_AG, CONFIRMED, LOST.
- rag_context: AI-extracted context from your menu, FAQs, and policies.
Reservations
Bi-directional Sync
Convert pre-qualified leads directly into reservations. The system checks real-time availability and pushes confirmed reservations to your existing platform (TheFork, CoverManager) bidirectionally.
/v2/reservations- Supports TheFork, CoverManager, Resy, OpenTable via native connectors.
- Real-time availability polling every 30 seconds.
- Atomic writes: reservation only confirms if the slot is available.
- Auto-sends confirmation WhatsApp/Email to the guest on success.
Webhooks
Event-Driven Architecture
Subscribe to real-time events from the LlamaChilly engine. Webhooks are signed with HMAC-SHA256 and retried up to 72 hours with exponential backoff on failure.
/v2/webhooks/subscribe- Signature header: X-LlamaChilly-Signature-256.
- Retry policy: 5 attempts, exponential backoff (1s → 32s → 8m).
- Maximum payload size: 512 KB.
- Delivery dashboard available in the sovereign control panel.
Event Hooks
Available Events
LlamaChilly emits structured events at every stage of the revenue pipeline. Build your custom logic on top of these atomic signals.
- lead.qualified: A lead has passed the AI scoring threshold.
- reservation.confirmed: Guest slot has been committed in the platform.
- cart.abandoned: Guest started but did not complete the booking flow.
- review.requested: Post-visit survey sent 7 days after dining.
- review.negative: Private alert triggered for sub-4 star ratings.
SDKs & Libraries
Official Client Libraries
Official SDKs are available for Node.js/TypeScript and Python. Both include full TypeScript definitions, retry logic, and native cluster routing.
- Node.js SDK: npm install @llamachilly/sdk. Supports CJS & ESM.
- Python SDK: pip install llamachilly. Compatible with Python 3.10+.
- Auto-retries on 429/503 with jitter.
- Built-in request tracing for observability stacks (OpenTelemetry).
// Install via npm
// npm install @llamachilly/sdk
import { LlamaChilly } from '@llamachilly/sdk';
const client = new LlamaChilly({
apiKey: process.env.LLAMA_API_KEY,
cluster: 'AMS-PDR-01',
version: 'v2',
});
console.log(client.status()); // => "READY"