cortyxia

The Memory Layer for AI Applications

Model-agnostic memory that unifies context, cuts costs 40-60%, and improves retrieval — with zero friction.

What Cortyxia Is

Cortyxia is a memory layer that sits between your application and any LLM provider (Anthropic, Google, DeepSeek, xAI, Llama, Groq, and more). It wraps around your existing API workflow and adds:

• Persistent memory — Every conversation enriches a shared knowledge base that compounds over time
• Zero-overhead capture — Use any AI exactly as you do today; context is automatically captured and structured at the infrastructure layer
• Precision retrieval — BM25 + semantic reranking surfaces exactly the memory your model needs, keeping prompts lean
• Model-agnostic — Switch providers instantly without changing your code
• Private memory keys — Every project gets isolated context that cannot cross-contaminate

Quick Start

npm install cortyxia

import { Cortyxia } from "cortyxia";

const client = new Cortyxia({
  isoUrl:  "https://app.cortyxia.com",
  isoToken: "iso-...",  // Your Cortyxia project token
});

// Automatic memory injection — zero code changes beyond this
const res = await client.chat.completions.create({
  messages: [{ role: "user", content: "What did we discuss yesterday?" }],
});

console.log(res.choices[0].message.content);

How It Works

1. You send a message through the SDK (or any HTTP client)
2. Cortyxia retrieves relevant memory from your project's knowledge base
3. Retrieved context is injected into the prompt
4. Request is forwarded to your configured provider
5. Response is captured and scored for future retrieval

Features

Cumulative Intelligence

Every resolved ticket, strategic decision, and customer conversation enriches your shared memory. New team members inherit years of knowledge on day one.

Token-Efficient Memory

Smart routing and semantic caching reduces total token usage 40-60%. Hot nodes rank higher, stale context gets deprioritized, and prompts stay lean.

Cross-Platform Sync

Real-time bidirectional synchronization keeps every connected app in lockstep. Native integrations with Salesforce, HubSpot, Zendesk, ServiceNow, Jira, Slack, Teams, and more.

Drop-in Compatibility

No refactoring required. Import, add your token, and keep using your existing code.

Configuration

const client = new Cortyxia({
  isoUrl:     "https://app.cortyxia.com",  // Required
  isoToken:   "iso-...",                     // Required — your project token
  timeout:    60000,                         // Optional — ms (default: 60000)
});

Advanced

Memory Seeding (Bulk Import)

await client.memory.add("User is a vegetarian", ["diet", "preference"]);

Direct Memory Query

const hits = await client.memory.query("vegetarian preferences", 5);

Links

• Website: https://www.cortyxia.com
• Documentation: https://docs.cortyxia.com
• GitHub: https://github.com/simar5244/CortyxiaAPI