The formal release memo for our new brand, website, APIs, SDKs and more
Today's a big day.
The announcement you're reading right now will start with the release of our new brand, IntegrationOS. It'll then detail why we're doubling down on building a real-time operating system for 3rd-party integration data and what it means for the future of computing.
To understand the significance of this announcement and the direction we're headed, you first have to understand the why.
Over the course of 2023, we met with hundreds of engineering leaders and developers building all types of products.
Quickly, a pattern emerged surrounding the process of building, testing and maintaining native integrations. The catalyst of extensibility for high-functioning product teams.
Integrations create pressure on software businesses.
They're mission-critical in unblocking revenue and product usage, but they're disliked by engineers tasked with building them.
Integrations are tedious and repetitive to build, which makes them resource heavy to launch. They're also consistently moving targets. APIs are constantly changing their authentication protocols, data models, documentation and other core parts of their platforms. The byproduct is a never ending maintenance cycle for engineers.
Today, a ton of engineering leaders are constantly finding themselves in a buy vs. build decision. They're deeply excited by the potential of what the new emerging category in Unified APIs can offer. But, they're consistently disappointed by the execution and the services of existing vendors in the space.
A Unified API sits in front of multiple backend APIs and handles the dirty work of authentication, data normalization, and request routing so you don’t have to. It aims to provide a simplified, consistent developer experience across disparate providers.
Today, there are a number of Unified API vendors. Almost all of them leverage the concept of a shared data schema and a normalized API.
The single API makes it really easy to turn on and off integration providers, which is why Unified APIs have grown in popularity. By integrating once, you get access to a catalog of dozens of normalized integration vendors, overnight.
The problem? Almost all vendors today are built on top of clunky ETL or sync-based architectures.
Most Unified APIs today leverage an Extract, Transform and Load (ETL) architecture under the hood.
They recursively hit and extract your users' 3rd-party integration data, transform it and load it into an internal database. They then expose an API for you to read and write transformed data via this internal database. ETL is great for certain use cases in data science, but create a ton of bloat when used for abstracting integrations.
One of the biggest issues with sync-based architectures is sync intervals, which cause a delay in data. It's common to see sync intervals of 4, 8 or even 24 hours, which is an extremely long time to wait for data for most use cases. It also creates a lot of complexity for product teams to deal with when designing the UX across integrations.
Sync-based architectures force you to trade off security and privacy for the efficiency normalization provides. They require you to adhere to the fact that all of your users' integration data will be stored in the Unified API vendor's database. All of it. They can't unify it unless it's extracted and loaded internally first.
The requirement for you to store your data means that if any of these vendors were to have security issues or a breach, all of your users' data would be exposed. An extremely high price to pay in one part of the business for efficiencies in another.
Sync-based architectures are not only slow and clunky, they cost a lot to run.
Today's Unified API vendors have to do a lot of unnecessary movement of your data. The redundancy leads to a ton of network fees and complicated infrastructure to manage. All of this complexity creates massive bloat, which they pass on to you, the customer, to pay for.
Let's look at a real world example
Merge is a Unified API vendor that uses a sync-based architecture.
According to Merge's public pricing plan, they charge $650 / mo for up to 10 Production Linked Accounts. A single QuickBooks integration will cost $3,250 / mo for just 50 customers ($650 * 5).
This is the most bare bones version of the product, has a 24-hour sync frequency for data and doesn't include any of the additional gated features you'll eventually need.
To see how this scales, multiply our projected cost for QuickBooks by all the integrations you'll need. Suddenly, you're looking at a $60,000-$100,000 / mo bill for a tool that was initially intended to save you time and money.
At IntegrationOS, we wanted to give CTOs and their teams all the benefits of a Unified API without the bloat.
To do so, we designed an API that uses a radically different real-time routing architecture and we built it in Rust.
Our API moves data from one point to another and transforms the data in-flight. By doing this, it completely eliminates the middleman and the need to load data in and out of an expensive intermediary. This simple, but unintuitive change, completely changes the category's economics.
The result is a light, efficient and highly reliable Unified API.
From a billing standpoint, it means we can offer a flexible Stripe-like pricing model, where our competitors use a fixed-rated Oracle-like pricing model.
Over the course of the next few weeks, we will be lifting the veil on a suite of new products, developer experiences and pricing plans.
We’re releasing the industry’s first real-time and truly Unified API.
When designing the API, we focused on a few very important product pillars:
Very soon, this API will be paired with a new type of LLM that'll be able to command itself. We're calling this new model IntegrationOS' Large Action Model (LAM).
The Unified API is complimented by a modern drop-in web component, called AuthKit.
AuthKit handles all the things frontend developers hate when tasked with urgently scaling out integrations for customers. It's purposefully built for React/Next.js with support for Svelte/SvelteKit and Nuxt/Vue coming soon.
Optimized user flows
One of the most time consuming pieces of building native integrations is building a seamless UI/UX that feels great for users and optimizes conversion. With AuthKit, you get both out of the box.
OAuth support
AuthKit supports OAuth 2. From the IntegrationOS Admin Dashboard, you can easily set up internal OAuth apps for any supported integration. This ensures your users stay within a branded experience and permissions stay within an environment you control.
Deploy-free customizability
AuthKit is built to strike the perfect balance of abstraction and customizability.
We've made it so that you're not overburdened with a ton of set up and controls. But, you are able to control the important things. Things like the visibility of integrations shown within AuthKit to your end users and the app theme (i.e., dark or light mode).
We're releasing a library of SDKs.
Together, they make getting IntegrationOS from a local machine to production really simple. Today's version of the AuthKit SDK already makes it really easy to add secure authentication to any React/Next.js frontend application.
The Unified SDK makes it easy to engage the Unified API when building web application features using properly permissioned and scoped access to your users' third-party integration data.
We're releasing a complete overhaul of our Admin Dashboard to make it simpler and faster to manage integrations at scale. The new interface comes with a visual canvas for viewing and organizing connected accounts, launching new integrations in real-time, managing API keys and more.
Recently, we’re starting to feel the negative effects of API fragmentation more and more with the evolution of LLMs.
Thanks to LLMs embedded within applications, it’s becoming easier for engineers (and non-engineers) to translate user intent into a specified action. But, we still have the limitation on the type of apps they can communicate with.
We're looking forward to working with developers building AI-enabled applications that need to communicate with thousands of 3rd-party integrations to both comprehend and convert user intent into action - both seamlessly and in real-time.
We're going to continue to scale out our integration ecosystem and listen to feedback on where we can improve both our core API resources, as well as the experiences wrapped around them.
We're aiming to have 200+ unified integrations in General Availability in the next few months. In the meantime, we'll continue to work with customers to uncover new cases and share them with our community.
