Getting Started

Learn how Xilos works, connect your first agent, and understand the core concepts of the platform.

Architecture at a Glance

Xilos operates as a middleware gateway between your AI agents and LLM providers. Every query passes through Xilos, where it is observed, secured, and orchestrated before reaching the destination model.

Agent → Xilos Gateway → LLM Provider
         │
         ├── Observe (log, track, cost)
         ├── Secure (guardrails, PII, restrictions)
         └── Orchestrate (route, cache, compress, enrich)

Quickstart

Create a Xilos Account

Get Your API Key

Navigate to Settings in the dashboard and copy your API key, or create a Virtual Key with budget limits.

Point Your Agent at Xilos

Set the base URL to https://api.xilos.ai/api/v1 and use your Xilos API key as the bearer token. Any OpenAI-compatible client works.

Send Your First Query

Make a chat completion request — Xilos will route it, log it, and apply any configured guardrails.

Verify in the Dashboard

Check the Query Log to see your query, the routing decision, governance actions, and cost.

Core Concepts

Xilos has a few fundamental building blocks you should understand:

Routing Rules — Direct specific types of queries to the most appropriate LLM based on intent, keywords, or AI-generated descriptions.
Restriction Rules — Security policies that inspect queries for sensitive content. Queries can be blocked, masked, or flagged.
Workflows — Multi-stage LLM pipelines that chain models sequentially or run them in parallel.
Context Compression — Reduces prompt token count by 50-90% before reaching the LLM.
Context Engine — Extracts atomic facts from responses and uses them to enrich future queries.
Semantic Caching — Caches LLM responses and reuses them for semantically similar queries.
Smart Routing — Uses an internal SLM to classify query intent and select the optimal model.

Learn more in Core Concepts.

Next Steps

Welcome Quickstart