Question 1

What is Horse Labs' LLM Gateway?

Accepted Answer

It's the infrastructure layer between your operation and the LLM providers. It routes each task to the best model — Claude, GPT, Gemini, or a local model — applies a budget ceiling per client, and records the cost of every operation.

Question 2

How does Horse Labs avoid AI provider lock-in?

Accepted Answer

Your operation talks to the gateway, not directly to the provider. Switching models becomes configuration, not migration — if one provider raises its price or goes down, another takes over without rewriting anything.

Question 3

Does the infrastructure run on my own company?

Accepted Answer

Yes. We provision on-premise or in your own VPC — AWS, Azure, or GCP — with Ansible and Terraform, isolated per tenant and under your own domain.

Question 4

How does per-client cost control work?

Accepted Answer

The model is multitenant: each client or cost center has its own budget, model, and usage trail. You track cost per operation — including over WhatsApp.

Question 5

Do I need to understand the technology to hire you?

Accepted Answer

No. In the AI as a Service model we deliver the finished result — report, automation, or agent — with the metric agreed before we start. You hire the result, not the tool.

Question 6

What's the difference between Horse Labs and Portkey or Helicone?

Accepted Answer

Portkey was acquired by Palo Alto Networks in May 2026 and folded into Prisma AIRS — an enterprise security product, with a roadmap outside the original team's control. Helicone was acquired by Mintlify and has been in maintenance mode since March 2026. Horse Labs is independent, with its own roadmap and a focus on organizational governance — not just observability. On top of that: unlimited logging with no charge per record volume, and self-hosted as a first-class citizen, not an enterprise afterthought.

Question 7

Is Horse Labs a platform or a consultancy?

Accepted Answer

Both, in stages: we deliver the infrastructure implementation (a service) and operate the platform that governs it continuously (a product). The client buys the implementation once and uses the platform indefinitely. For those who already have a technical team: you can implement it in-house using our docs and use just the governance layer as SaaS.

Investment structure

AI as Infrastructure

AI as a Service