How Enata Built Production AI Infrastructure Without the Overhead

Meet Enata

Enata is an early-stage startup building in the AI space. Like many startups, they need to move fast with limited resources. Smit Shah, who previously worked with Prefect, knew the importance of reliable infrastructure—and knew that building it from scratch wasn't where a small team should spend their time.

The Startup Reality

"We are a startup. We are a few people and so we need to be scrappy," Smit explained. The constraint shaped their requirements: "Given that we are small and scrappy, we want tools to be serverless, inherently working. That's kind of the trend with this new set of developer tools."

Serverless wasn't just a preference—it was a necessity. A small team can't afford to dedicate engineers to infrastructure management.

The Best Onboarding Experience

What set Horizon apart was how quickly the entire team could get productive. "We found that Horizon actually had the best onboarding experience on my team," Smit said.

The accessibility surprised him. "Even people who have not done tech before were able to onboard onto the tool, which is a very strong sign of ease of use."

For a startup where everyone wears multiple hats, this matters. Tools that require deep technical expertise become bottlenecks. Tools that anyone can use become force multipliers.

Production Guarantees Without the Overhead

Beyond ease of use, Horizon delivered the reliability Enata needed to serve their customers. "It also had very good guarantees that we expected to actually provide to our customers."

The deployment workflow was seamless: "Just use a GitHub branch. It automatically, seamlessly deploys to a preview environment where you could point your production to."

No YAML wrestling. No container configuration. Push code, get a deployed server.

Trust Built on Track Record

Smit's confidence in Horizon came partly from his experience with Prefect. "Obviously since Prefect, I know that you guys are very reliable. I pushed for Prefect here as well. I was like, 'Yeah, this thing works. I know this thing works.'"

That trust translated directly to Horizon: "We have been using it a lot."

The Alternative: Wrangling Kubernetes

Smit was clear about what Horizon replaced. "You could wrangle with Kubernetes and autoscaling, but you need a lot more expertise as well as a lot more manpower."

For a small team, the math is simple. "Horizon is a very robust tool. We need tools like that at higher and higher level abstractions to make the entire ecosystem of agentic orchestration with high reliability and guardrails possible."

Building for the Real World

Smit sees Horizon as part of a new generation of tools. "I see Horizon as an AI-native, next-generation tool stack—similar to Supabase or Vercel."

The comparison is telling. Just as Vercel abstracts away frontend deployment and Supabase abstracts away backend infrastructure, Horizon abstracts away MCP server deployment.

This abstraction matters because real-world AI workflows are complex. "Real-world workflows are a lot more messy and multi-step and have a lot more nuance than a demo."

Production AI needs production infrastructure. Horizon provides that without requiring a production infrastructure team.

Key Results

Onboarded non-technical team members: Even team members without engineering backgrounds could use Horizon
Achieved production reliability: Serverless architecture with guarantees suitable for customer-facing applications
Eliminated infrastructure overhead: GitHub-based deployment with automatic preview environments
Reduced expertise requirements: No need for Kubernetes or autoscaling knowledge

Lessons for Lean Teams

Enata's experience reflects a broader truth: startups can't afford to build commodity infrastructure. The teams that win are the ones that outsource undifferentiated heavy lifting to platforms built for the job.

MCP server deployment is exactly that kind of infrastructure—essential but undifferentiated. Horizon lets small teams focus on what makes their product unique while standing on production-grade foundations.

Try Horizon Today

Building AI infrastructure with a small team? Get started with Horizon and deploy production-ready MCP servers without the overhead.