# TrueFoundry

> Enterprise-ready AI Gateway and agentic deployment platform for governing, deploying, scaling, and tracing AI workloads across VPC, on-prem, hybrid, or public cloud environments.

TrueFoundry is an enterprise-grade AI Gateway and agentic deployment platform that enables organizations to govern, deploy, scale, and trace AI workloads with full security and compliance. It provides a unified control layer for managing LLMs, MCP servers, AI agents, and model fine-tuning across any infrastructure — on-prem, VPC, air-gapped, or multi-cloud. Named in the Gartner® 2025 Market Guide for AI Gateways, TrueFoundry is trusted by teams at NVIDIA, ResMed, Whatfix, Innovaccer, and others to accelerate AI production timelines while reducing infrastructure costs.

- **AI Gateway** — *Centralize LLM access with universal API, virtual models, RBAC, semantic caching, weight/latency/priority-based routing, fallbacks, rate limiting, and budget controls across all model providers.*
- **MCP Gateway** — *Register, discover, and govern MCP servers with schema validation, access control, metrics, and support for advanced authentication and self-hosted MCPs.*
- **Prompt Lifecycle Management** — *Version, manage, and monitor prompts with guardrails and partner integrations to ensure repeatable, high-quality agent behavior.*
- **AI Deploy Platform** — *Host any LLM or custom model using vLLM, TGI, or Triton backends; deploy agents built with LangGraph, CrewAI, AutoGen, or custom frameworks in fully containerized, production-ready environments.*
- **Training & Fine-Tuning** — *Launch fine-tuning jobs on your own data, track experiments, and push updated checkpoints directly to production in one unified flow.*
- **Full Agent Observability** — *Trace every step from prompt to tool/model execution with metrics, latency, and outcomes; integrates with Grafana, Datadog, Prometheus via OpenTelemetry.*
- **GPU Orchestration & Autoscaling** — *Automatically schedule and scale GPU workloads, support NVIDIA MIG and time slicing for fractional GPU sharing, and continuously rightsize infrastructure to reduce cloud waste.*
- **Enterprise Security & Compliance** — *SOC 2, HIPAA, and GDPR compliant with SSO, granular RBAC, immutable audit logging, real-time policy enforcement, and org-level multi-tenant management.*
- **Flexible Deployment Modes** — *Deploy as SaaS, VPC/on-prem, air-gapped, or hybrid; data never leaves your domain, ensuring complete sovereignty.*
- **Integrations** — *Framework-agnostic support for LangGraph, CrewAI, AutoGen, vLLM, TGI, Triton, Grafana, Datadog, Prometheus, and more.*

## Features
- AI Gateway with universal API
- MCP Gateway and agents registry
- Prompt lifecycle management
- Model hosting with vLLM, TGI, Triton
- Training and fine-tuning
- Agent deployment (LangGraph, CrewAI, AutoGen)
- Full agent observability and tracing
- GPU orchestration and autoscaling
- Fractional GPU support (MIG and time slicing)
- RBAC and SSO
- Immutable audit logging
- SOC 2, HIPAA, GDPR compliance
- Semantic and simple caching
- Weight/latency/priority-based routing
- Fallbacks and rate limiting
- Budget limiting
- VPC, on-prem, air-gapped deployment
- OpenTelemetry integration
- Multi-tenant org management
- Real-time policy enforcement

## Integrations
LangGraph, CrewAI, AutoGen, vLLM, TGI, Triton, Grafana, Datadog, Prometheus, OpenTelemetry, AWS, GCP, Azure, Kubernetes, OpenAI, Anthropic, Hugging Face

## Platforms
WEB, API

## Pricing
Freemium — Free tier available with paid upgrades

## Links
- Website: https://www.truefoundry.com
- Documentation: https://docs.truefoundry.com
- Repository: https://github.com/truefoundry
- EveryDev.ai: https://www.everydev.ai/tools/truefoundry