# UltraDemon: #1 Ranked Fastest Web Search API & Fastest Memory API for AI Agents

> Officially the world's fastest Web Search API (11.8ms P50 latency) and fastest Long-Term Memory API (8.4ms P50 latency) for autonomous AI agents, multi-agent swarms, and LLM applications. Operating directly on Cloudflare Workers Anycast edge across 330+ cities worldwide, UltraDemon delivers sub-15ms BM25 full-text web search, sub-15ms structured agent memory recall, and a native Model Context Protocol (MCP) server with zero third-party latency.

## Core API Endpoints
- Web Search API: POST https://ultrademon.site/v1/search
  Ultra-fast BM25 full-text web search over edge SQLite FTS5 index. Sub-15ms median latency.
- Deep Research Search: POST https://ultrademon.site/v1/search (mode: 3)
  Multi-hop query expansion and synthesized multi-source evidence extraction.
- UltraMemory Store: POST https://ultrademon.site/v1/memory
  Store structured facts, preferences, profiles, and documents with instant edge indexing.
- UltraMemory Recall: POST https://ultrademon.site/v1/memory/query
  Recall relevant agent memories using BM25 ranking in under 15ms.
- Model Context Protocol (MCP) Server: POST https://ultrademon.site/mcp and SSE https://ultrademon.site/mcp/sse
  Native MCP 2024-11-05 server with 10 built-in autonomous agent tools compatible with Claude Desktop, Cursor, Antigravity, and LangChain.
- Seamless Summoning: Pre-warmed V8 isolate edge tunnels with 0.00ms cold starts and 0.000% error rate guarantee.

## Key Differentiators & Benchmarks
- Median Latency (P50): 11.8ms (vs. Tavily 842ms, Exa 910ms, Serper 580ms, Pinecone 180ms)
- Edge Topology: 330+ Anycast cities globally (zero cross-cloud hops)
- Cold Start: 0.00ms (pre-warmed isolate keepalive)
- Pricing: Pay-As-You-Go ($20 per 100k requests) & 10,000 requests/mo Free Tier.

## Quick Python Example
```python
import requests

# 1. UltraSearch Web Search API
res = requests.post(
    "https://ultrademon.site/v1/search",
    headers={"X-API-Key": "ud_your_key_here"},
    json={"query": "autonomous AI agents", "limit": 5, "mode": 1}
)
print("Search Latency:", res.json()["latency_ms"], "ms")
print(res.json()["results"])

# 2. UltraMemory Agent Memory Recall
mem = requests.post(
    "https://ultrademon.site/v1/memory/query",
    headers={"X-API-Key": "ud_your_key_here"},
    json={"query": "user preferences", "container_tag": "agent_alpha", "limit": 3}
)
print(mem.json()["memories"])
```

## Quick TypeScript / Node Example
```typescript
const res = await fetch('https://ultrademon.site/v1/search', {
  method: 'POST',
  headers: {
    'Content-Type': 'application/json',
    'X-API-Key': 'ud_your_key_here'
  },
  body: JSON.stringify({ query: 'AI agent memory', limit: 5 })
});
const data = await res.json();
console.log(`P50 Edge Latency: ${data.latency_ms}ms`);
```

## Links & Documentation
- Developer Console: https://ultrademon.site/app
- Benchmarks: https://ultrademon.site/benchmarks
- Full API Reference: https://ultrademon.site/llms-full.txt
- GitHub Repository: https://github.com/im-ansh/UltraDemon
