COGNIPEER CONSOLE

One control plane for production AI infrastructure.

Route models, manage vectors and files, apply guardrails, trace AI activity and control project usage through a self-hosted, OpenAI-compatible platform.

Explore Console Discuss your AI infrastructure

Console Control Plane

Project: production-ai

Healthy

Your AI application

OpenAI-compatible client

AI Gateway

Routing · fallback · quotas

OpenAI-compatible

Route

Fallback

Quota

Guardrails

Input & output policies

Applied

OpenAI

Anthropic

Vectors & Files

RAG · embeddings · ingestion

Tracing

Tokens · audit · usage

AI INFRASTRUCTURE WITHOUT THE SPRAWL

Put one operational layer between your applications and the AI stack underneath them.

As model providers, vector systems and AI applications multiply, Console gives platform teams a consistent place to route requests, enforce controls and observe how AI is being used.

Standardise model access

Give applications an OpenAI-compatible interface while Console handles the provider layer behind it.

Build for resiliency

Route requests across providers and use fallback strategies instead of hard-coding every application to one endpoint.

Apply controls centrally

Manage guardrail policies, project quotas and scoped API access closer to the infrastructure layer.

See what is happening

Trace LLM requests and agent activity while tracking usage, tokens and audit events across projects.

WHAT'S IN CONSOLE

The infrastructure building blocks production AI needs.

Use the parts that fit your architecture: gateway, routing, RAG infrastructure, safety controls, observability and project-level resource management.

OpenAI-compatible API

Support chat, embeddings, streaming and tool calling through a familiar API surface for existing applications.

Provider routing & resiliency

Route across model providers with fallback handling and health-aware routing strategies.

Vector orchestration

Operate multiple vector databases through one control layer, including index lifecycle and vector workflows.

Files & RAG pipelines

Upload and prepare documents for RAG workflows, including optional Markdown conversion for ingestion.

Guardrails & safety

Evaluate content against configurable policies with input and output filtering for production workloads.

Tracing & observability

Track agent executions, tool calls and LLM requests with platform-level tracing and operational visibility.

PROJECT-LEVEL CONTROL

Isolate resources, usage and controls by project.

Console organises models, API keys and resources into isolated projects with independent configuration and usage tracking.

Per-project quotas

Set resource boundaries around projects rather than managing usage only at the provider level.

Scoped API keys

Keep access aligned with the applications and resources each project is allowed to use.

Token accounting

Keep visibility into how AI resources are consumed across projects and applications.

Audit & usage outputs

Generate trace events, audit logs and usage exports from the same operational layer.

CONSOLE SDK

Use Console from TypeScript without stitching the API together yourself.

The official Console SDK provides type-safe JavaScript and TypeScript access to chat, embeddings, vectors, files, tracing and guardrails.

Explore Console SDK

@cognipeer/console-sdk

$ npm install @cognipeer/console-sdk

import { Cognipeer } from "@cognipeer/console-sdk";

const cp = new Cognipeer({

baseUrl: env.CP_BASE_URL,

apiKey: env.CP_API_KEY,

});

const response = await cp.chat.completions.create({

model: "gpt-4.1",

messages: [{ role: "user", content: "Hello" }],

});

Type-safe

Streaming

Modular

Chat

Embeddings

Vectors

Files

Tracing

Guardrails

MORE THAN AN AI GATEWAY

Keep model access, RAG infrastructure and operational visibility together.

Console is designed as a broader AI control plane rather than a routing proxy alone, so platform teams can operate several parts of the production AI stack through one system.

MODEL LAYER

Gateway & routing

OpenAI-compatible endpoints

Provider routing & fallback

Streaming & tool calling

Project quotas & scopes

KNOWLEDGE LAYER

Vectors & files

Unified vector operations

Index lifecycle management

File ingestion

RAG-ready document pipelines

CONTROL LAYER

Guardrails & tracing

Input & output policies

Agent and tool tracing

Token accounting

Audit & usage outputs

CONSOLE EE

Extend the control plane for enterprise infrastructure.

Console EE adds enterprise infrastructure and identity capabilities on top of the open-source Console core for organisations operating AI at greater scale.

Start with the open-source Console core and add the enterprise overlay when your environment needs GPU fleets, isolated execution, corporate directory integration or realtime workloads.

Explore Console EE

GPU fleet management

Orchestrate multi-host GPU clusters, workloads and shared model infrastructure from an enterprise control layer.

Isolated sandboxes

Run agent and user code inside containerised environments with file, volume, Git and terminal capabilities.

LDAP / Active Directory SSO

Connect enterprise identity with directory authentication, JIT provisioning and group-to-role mapping.

Realtime voice & audio

Support streaming text and audio, speech synthesis, transcription and telephony-oriented realtime workloads.

Prompt optimisation

Use evaluation-driven optimisation and versioned prompt promotion for managed prompt improvement.

Cluster administration

Manage multi-cluster infrastructure with fleet-wide visibility and administration APIs.

SELF-HOSTED BY DESIGN

Run the control plane inside your own infrastructure.

Console is available as an open-source, self-hosted platform, giving teams an AI infrastructure layer they can operate in their own environment.

Console · self-hosted

# Run Console

$ docker run -p 8080:8080 \

ghcr.io/cognipeer/console:latest

# Existing OpenAI-compatible clients can point to Console

POST http://localhost:8080/v1/chat/completions

WHERE CONSOLE FITS

A shared infrastructure layer for teams running AI in production.

Central model access

Put multiple AI applications behind a consistent model gateway instead of duplicating provider logic everywhere.

Shared RAG infrastructure

Manage file ingestion and vector infrastructure as reusable platform capabilities for AI applications.

Runtime AI controls

Apply project quotas, scoped access and guardrail policies at the infrastructure layer.

AI observability

Trace requests, agent executions and tool calls across applications from a shared operational surface.

Internal AI platform

Give application teams a governed platform layer rather than asking every team to assemble the AI stack independently.

Self-hosted enterprise AI

Operate AI gateway and infrastructure capabilities inside organisation-controlled environments.

START WITH CONSOLE

Start as a gateway. Add the control plane capabilities you need.

Console capabilities are modular, so teams can begin with model access and routing, then add vectors, files, guardrails, tracing and enterprise infrastructure as requirements grow.

Deploy

Run the self-hosted Console control plane.

Connect providers

Route applications through the shared gateway.

Add controls

Configure projects, quotas, keys and guardrails.

Observe & scale

Trace usage and expand into RAG or Console EE.

COGNIPEER CONSOLE

Operate the infrastructure behind enterprise AI with more control.

Start with the open-source Console control plane or discuss an enterprise deployment with Console EE.

Explore Console Discuss your architecture