Why developers deserve better than the services from AWS

To understand why developers deserve better, we have to understand the history of AWS

Amazon Web Services (AWS) evolved from Amazon's internal infrastructure efforts into a global cloud computing leader. The journey began in the early 2000s when Amazon sought to improve its IT infrastructure

In 2000, Amazon started working on making its infrastructure more reliable and scalable

Jeff Bezos envisioned a platform where businesses could rent computing power

By 2003, Andy Jassy had taken over the project and mapped out the vision for an "Internet OS" made up of foundational infrastructure primitives

The first public AWS service, Amazon Simple Queue Service (SQS), launched in 2006

This was quickly followed by the introduction of Amazon Simple Storage Service (S3) and Elastic Compute Cloud (EC2), which formed the core of AWS's initial offerings

AWS built the foundation of the internet with the idea of renting compute

Today they're still the #1 player in the market for renting compute like GPUs

Fast forward to 2015, AWS released Amazon Machine Learning followed by services like Textract in 2019

In 2013 AWS launched its AWS activate program giving startups credits making it cost-effective to build their startups on top of AWS and lock in over the years!

Today we have two products that are startup favorites that devs prefer over AWS

• Vercel (Launched 2015) - Frontend hosting and more

• Supabase (Launched 2019) - Database hosting and more

Fun fact - both platforms are built on top of AWS compute

What makes both Vercel & Supabase special is the ease of use at any scale by understanding their audiences: developers

What I'm trying to get to is that AWS is amazing at computing, the best at building the raw foundation but they are horrible at building services for developers

That goes for Azure and GCP too!

Large cloud providers -> Great compute

Service providers -> Great experience

Now in the age of AI, we see more applications using AI in ways we can't imagine.

Cloud providers have two suite of services for AI:

• Model hosting layer with their GPU services (AWS bedrock, GCP VertexAI, Azure AI Model Catalog, etc)

• Custom models as services (AWS Textract, Google Translate, Azure Speech-to-text, etc)

Well again, they pretty much aren't that great at providing the service but amazing at providing the compute infra

Multiple providers are trying to solve model hosting and making it super simple: Baseten, HuggingFace, Replicate, etc, and the list goes on...

We want to solve custom models by providing the best suite of custom models packaged into a single SDK for developers to plug and play in their tech stack!

AWS models are on average 6 years old and difficult to access. They're also expensive at scale for the quality provided

We solve this by building and training our models that beat both quality and lower cost compared to the big 3 cloud providers

We make DX so easy, all you need is an API key and no setup to use any model at scale

Check out JigsawStack and let us know what you think! We have a long way to go and tons of improvements to be made