TrueFoundry AI Gateway

TrueFoundry AI Gateway

Connect, observe & control LLMs, MCPs, Guardrails & Prompts

2.1K followers

TrueFoundry’s AI Gateway is the production-ready, control plane to experiment with, monitor and govern your agents. Experiment with connecting all agent components together (Models, MCP, Guardrails, Prompts & Agents) in the playground. Maintain complete visibility over responses with traces and health metrics. Govern by setting up rules/limits on request volumes, cost, response content (Guardrails) and more. Being used in production for 1000s of agents by multiple F100 companies!
Interactive
TrueFoundry AI Gateway gallery image
TrueFoundry AI Gateway gallery image
TrueFoundry AI Gateway gallery image
TrueFoundry AI Gateway gallery image
TrueFoundry AI Gateway gallery image
TrueFoundry AI Gateway gallery image
TrueFoundry AI Gateway gallery image
TrueFoundry AI Gateway gallery image
Free Options
Launch Team / Built With
Auth0
Auth0
Start building with Auth0 for AI Agents, now generally available.
Promoted

What do you think? …

Anuraag Gutgutia

Hey Product Hunt, Anuraag here, co-founder at TrueFoundry 👋

When we first thought about a “gateway”, we imagined a simple LLM routing layer in front of models. Pick a model, send traffic, switch if needed. Easy… or so we thought.

Once teams started putting agents and MCPs into production, we realised the hard stuff wasn’t just about routing. It is:

  • Different MCP auth flows for every internal system.

  • Traces & logs that break once you chain models, tools, and agents

  • Data residency and “this data must stay in this region” rules,

  • Security asking “who called what, when, with which payload?”,

  • Product teams need to swap models without rewriting everything.

So the “router” slowly turned into a proper control plane that sits between your apps, LLMs, and MCPs - making sure traffic is reliable, auditable, compliant, and still fast for developers to ship on.

Today, TrueFoundry’s AI Gateway sits at the center of production traffic across 10+ Fortune 500s, powering their internal copilots and agents while platform teams use it to keep costs, safety, and observability under control - rather than maintaining a pile of custom glue code.

🔗 Sign Up Link: Please try and give us feedback! 🙏
🎁 Launch perk: 3‑month free trial for the PH community


If you’re wrestling with MCP auth, logging, or data policies, drop your setup in the comments - curious to see how you are wiring your stack today!

Aamir Siddiqui

@agutgutia Love it. Identity is the new control plane for agentic AI. this gateway is glueing all components in a seamless way for enterprise use.

Anuraag Gutgutia
@aamir_siddiqui2 hope you are doing great progress on the test automation tooling as well.
Viktor Shumylo

That’s a strong value proposition. A unified control plane with real governance, cost controls, and full traceability is exactly what most teams lack when scaling agents past prototypes. The fact that it’s already running thousands of agents in production for F100 companies adds real credibility.

Anuraag Gutgutia

@vik_sh We’re incredibly excited about the future of AI gateways - one unified control plane that seamlessly manages models, MCPs, and agents. We believe this will become one of the most critical layers of modern AI infrastructure.

Shrey Shrivastava

It seems useful from security and monitoring perspective, the fact that we get detailed traces could help. I wonder what all gets traced here?

Anuraag Gutgutia

@shrey_shrivastava We provide very detailed traces with LLM, MCP, Guardrails, etc invocation

Nikunj Bajaj

@shrey_shrivastava  Basically, AI Gateway is the entry point to all components like - models, MCPs, Guardrails & Prompts and all of those get traced here. Some examples,

  • If a request went to the LLM, ended up invoking an MCP server, and then implement a guardrail - you will see all of them logged in the traces.

  • You get to see your prompts, completions, tool call results, latency, time to first token etc.

  • If one of your model fallback or rate-limiting rules got applied - that also gets captured in the traces and these will also be flagged separately.

Besides these, you can also monitor aggregate metrics at request level, MCP level, guardrail level where you can observe costs, error rates, overall latency and you can dissect these by users, models, or any custom metadata as well.

Please provide feedback if you ideally wanted something else to be logged / traced that we are not covering here. One of the major goal of this launch with us is to collect feedback on the product! :)

Daniel Agoston

Who is your main competitor and how does TrueFoundry AI Gateway differentiate itself from them?

Anuraag Gutgutia

@danagoston For competitors, we usually encounter incumbent API gateways or internal, in-house gateway solutions. They start out simple but quickly become complex to build and maintain - especially when it comes to data residency, advanced configurations, and overall time to market.

Chilarai M

Great launch! The AI Gateway brings all the moving parts of agents into a single, production-ready control plane. As someone building Swytchcode for API workflows, I love seeing infra like this push the ecosystem forward.

Excited to see this climb the charts!

Anuraag Gutgutia

@chilarai Thank you for your support! Swytchcode and Truefoundry are both critical pieces of infrastructure for today's time.

Chilarai M

@agutgutia Thanks a lot. Looking forward to connecting with you.
Go for the win!!

Tymofii Pidlisnyi

Looks solid. We’re building agentic workflows at Evatar, so I’m definitely curious to see how your Gateway handles guardrails and multi-model routing in real production. When are you planning to open it up for broader access?

Anuraag Gutgutia

@tima_fey Thanks! Definitely a lot of synergies here. Gateway is a very critical piece while putting agentic flows in production. We are opening up next week :)

Siful

Congratulations on the lunch. Best wishes for the team TrueFoundry

Anuraag Gutgutia

@getsiful Thanks for your support! Would love for the team to try out the product post launch.

123
•••
Next
Last