Back to NI-Stack Pitch Deck

QuickStart & Benchmark Guide

For Technical Due Diligence: Prove the 29% token savings and 0.5ms AEGIS latency using your own internal integration tests.

1

The 1-Line Drop-in Replacement

The NI-Stack requires zero architectural changes to your existing application. It acts as a transparent, drop-in proxy. Simply point your OpenAI or Anthropic client to the DESTILL endpoint.

// ❌ Old Code (Direct to OpenAI)
const client = new OpenAI({
  baseURL: 'https://api.openai.com/v1',
  apiKey: process.env.OPENAI_API_KEY
});

// ✅ New Code (Routed through NI-Stack)
const client = new OpenAI({
  baseURL: 'https://api.destill.ai/v1',
  apiKey: process.env.DESTILL_API_KEY
});

Note: You configure your OpenAI/Anthropic keys within the DESTILL secure dashboard. The NI-Stack handles the API pass-through securely.

2

Observe Payload Compression (STENO & ORACLE)

Once integrated, inspect the network payloads. You will immediately notice the effects of STENO (Shorthand Compression) and ORACLE (Hash Pointer Memory).

Without NI-Stack (748 Tokens)
{
  "messages": [
    {"role": "user", "content": "Hello, my name is John and my order number is 12345."},
    {"role": "assistant", "content": "Hi John, how can I help you?"},
    {"role": "user", "content": "Where is my order?"}
  ]
}
With NI-Stack (12 Tokens)
{
  "messages": [
    {"role": "user", "content": "Where is my order?"}
  ],
  "ni_oracle_context": "blake3:4a5b6c7d..."
}
3

Verify Technical Benchmarks

Run your internal load tests. You should observe the following verified metrics derived from the 16M prompt dataset running on the Hetzner AX42 production environment.

0.5ms
AEGIS Cascade Latency
29.4%
Avg Token Reduction
99.999%
Threat Blocking (TPR)