Our Best Customers Are Now Robots

It’s weird to say this out loud! For years, one of our calling cards was “developer experience”. We made a decision, early on, to be a CLI-first company, and put a lot effort into making that CLI seamless. For a good chunk of our users, it really is

Fly.io logo
Fly.io
•
Kurt Mackey
•10 min read•beginner•
--
•View Original
AWSAWS EC2AWS LambdaCloudflare WorkersDockerHerokuOAuth

Overview

The article discusses the shift in user demographics on Fly.io's platform, highlighting how robots have become the primary drivers of growth. It explores the implications of this trend on developer experience (DX) and the need to adapt to the requirements of automated systems.

What You'll Learn

1

How to leverage Fly Machines for quick deployment of applications

2

Why modern robots require different cloud infrastructure compared to human users

3

How to implement storage solutions that cater to LLM workflows

Prerequisites & Requirements

  • Understanding of Docker and cloud computing concepts
  • Familiarity with command-line interfaces(optional)

Key Questions Answered

What are the primary drivers of growth on Fly.io's platform?
The primary drivers of growth on Fly.io's platform are robots, which have become the main users, overshadowing human developers. This shift indicates a need for cloud services to adapt to the requirements of automated systems, focusing on features that benefit these users.
How do Fly Machines differ from AWS EC2 and Lambda?
Fly Machines are Docker containers running as hardware virtual machines that can start quickly, similar to AWS Lambda, but they can also persist for as long as needed, unlike Lambda's short-lived nature. This flexibility allows for both server and batch job deployments.
What storage solutions are recommended for LLM workflows?
For LLM workflows, a filesystem and object storage are recommended to accommodate stateful trial-and-error iterations that LLMs require. This contrasts with traditional immutable container workflows that rely on static OCI containers.
What networking features does Fly.io provide for robots?
Fly.io provides an automatically connected load-balancing Anycast network with TLS, which is essential for robots interfacing with external APIs. This includes support for long-lived connections necessary for protocols like MCP used by LLMs.

Technologies & Tools

Some links below are affiliate links. We may earn a commission if you make a purchase.

Containerization
Docker
Used to run applications as Fly Machines on Fly.io.
Database
Postgresql
Mentioned as a managed service for data storage.

Key Actionable Insights

1
Consider adapting your cloud infrastructure to better serve automated systems, such as robots and LLMs.
As robots become a significant user base, understanding their needs can help optimize your platform for better performance and user satisfaction.
2
Utilize Fly Machines for rapid application deployment and management.
The ability to quickly start and stop Fly Machines allows for flexible development and testing, making it easier to manage resources efficiently.
3
Implement a filesystem for LLM workflows to facilitate iterative development.
This approach allows LLMs to manage state effectively, which is crucial for their operation and performance.

Common Pitfalls

1
Failing to recognize the differences in requirements between human users and automated systems can lead to suboptimal platform performance.
This oversight can result in a lack of features that cater specifically to robots, which could hinder growth and user satisfaction.

Related Concepts

Cloud Computing
Containerization
Devops Practices
Machine Learning Workflows