Overview - co-mind.ai

The co-mind.ai — Private AI Platform provides enterprise-grade access to multiple AI models through a unified, OpenAI-compatible API. Deploy on your own infrastructure with full data sovereignty.

Key Features

Multimodal Support

Process text, images, and audio inputs through a single API.

OpenAI Compatibility

Drop-in replacement for OpenAI API endpoints — use existing SDKs and tools.

Multiple AI Backends

Access vLLM, Ollama, llama.cpp, SambaNova, and more through one interface.

Real-time Streaming

Stream responses via Server-Sent Events for better user experience.

Function/Tool Calling

Extend AI capabilities with custom tools and function calls.

Security Sanitizer

Built-in injection/jailbreak detection, PII redaction, and configurable policies.

Platform Services

Core AI

Service	Description
Chat Completions	OpenAI-compatible chat with streaming, vision, and tool calling
Text Completions	Prompt-based text generation
Embeddings	Vector embeddings for semantic search and similarity

Knowledge Base (RAG)

Service	Description
Knowledge Bases	Upload and manage document collections
Stateless Chat	Chat with knowledge bases — you manage conversation history
Stateful Sessions	Server-managed conversation history with KB context
Context Retrieval	Query KBs directly for relevant document chunks

Echo Engine (Audio)

Service	Description
Transcription (STT)	Synchronous and async audio transcription with speaker diarization
Text-to-Speech (TTS)	Convert text to speech with multiple voices and languages
Recordings	Manage audio recording metadata and link to transcription jobs

Research & Analysis

Service	Description
Researcher	Multi-provider web search, content scraping, deep research with iterative reasoning
Document Analyzer	Extract structured data from documents with human-in-the-loop review
Grant Proposals	AI-assisted grant proposal generation

Administration

Service	Description
Tenant Management	Multi-tenant isolation with sub-organizations
Directory Integration	LDAP/Active Directory and Microsoft Entra ID (Azure AD)
Security Sanitizer	Configurable content safety policies
Audit Logs	Comprehensive audit trail with filters and aggregation

Supported Models

The platform is model-agnostic — deploy and serve any model through the supported backends. There are no restrictions on which models you can use.

Supported Backends

Backend	Description	Example Models
vLLM	High-performance GPU inference	Llama, Falcon, Qwen, Mistral, Gemma, and any HuggingFace model
Ollama	Local model server	Any model from the Ollama library
llama.cpp	CPU/GPU inference with GGUF models	Any GGUF-format model
Infercom	EU sovereign AI inference (SambaNova hardware)	Llama, DeepSeek, Qwen, and other hosted models
OpenAI	OpenAI API passthrough	GPT-4o, GPT-4, GPT-3.5, and all OpenAI models
Anthropic	Anthropic API passthrough	Claude 4, Claude 3.5, and all Anthropic models

Use GET /v1/models to list all available models on your instance and GET /v1/capabilities to check which features each backend supports.

Base URL

Your specific API endpoint URL will be provided with your credentials. The general format is:

https://your-comind-instance.example.com

All API endpoints are prefixed with /v1/.

Next Steps

Quickstart

Make your first API call in minutes.

Authentication

Set up JWT or PAT authentication.

​Key Features

Multimodal Support

OpenAI Compatibility

Multiple AI Backends

Real-time Streaming

Function/Tool Calling

Security Sanitizer

​Platform Services

​Core AI

​Knowledge Base (RAG)

​Echo Engine (Audio)

​Research & Analysis

​Administration

​Supported Models

​Supported Backends

​Base URL

​Next Steps

Quickstart

Authentication

Key Features

Platform Services

Core AI

Knowledge Base (RAG)

Echo Engine (Audio)

Research & Analysis

Administration

Supported Models

Supported Backends

Base URL

Next Steps