Why Knit

TL;DR

If you are exploring Unified APIs or Embedded iPaaS solutions to scale your integrations offerings, evaluate them closely on two aspects - API coverage and developer efficiency. While Unified API solutions hold great promise to reduce developer effort, they struggle to provide 100% API coverage within the APPs they support, which limits the use cases you can build with them. On the other hand, embedded iPaaS tools offer great API coverage, but expect developers to spend time in API discovery for each tool and build and maintain separate integrations for each, requiring a lot more effort from your developers than Unified APIs.

Knit’s AI driven integrations agent combines the best of both worlds to offer 100% API coverage while still expecting no effort from developers in API discovery and building and maintaining separate integrations for each tool.

Let’s dive in.

Solutions for embedded integrations

Hi there! Welcome to Knit - one of the top ranked integrations platforms out there (as per G2).

Just to set some context, we are an embedded integration platform. We offer a white labelled solution which SaaS companies can embed into their SaaS product to scale the integrations they offer to their customers out of the box.

The embedded integrations space started over the past 3-4 years, and today, is settling down into two kinds of solutions - Unified APIs and Embedded iPaaS Tools.

You might have been researching solutions in this space, and already know what both solutions are, but for the uninitiated, here’s a (very) brief download.

Unified APIs help organisations deliver a high number of category-specific integrations to market quickly and are most useful for standardised integrations applicable across most customers of the organisation. For Example: I want to offer all my customers the ability to connect their CRM of choice (Salesforce, HubSpot, Pipedrive, etc.) to access all their customer information in my product.

Embedded iPaaS solutions are embedded workflow automation tools. These cater to helping organisations deliver one integration at a time and are most useful for bespoke automations built at a customer level. For Example: I want to offer one of my customers the ability to connect their Salesforce CRM to our product for their specific, unique needs.

Knit started its life as a Unified API player, and as we spoke to hundreds of SaaS companies of all sizes, we realised that both the currently popular approaches make some tradeoffs which either put limitations on the use cases you can solve with them or fall short on your expectations of saving engineering time in building and maintaining integrations.

But before we get to the tradeoffs, what exactly should you be looking for when evaluating an embedded integration solution?

While there will of course be nuances like data security, authentication management, ability to filter data, data scopes, etc. the three key aspects which top the list of our customers are:

1. Whether the solution covers the APP you want to integrate with
2. Whether the solution covers the APIs within those APPs to solve for YOUR use case, and
3. How much effort does it take for YOUR developers to build and maintain integrations on the platform

Now let’s try and understand the tradeoffs which current solutions take and their impact on the three aspects above.

The Coverage problem with Unified APIs

The idea of providing a single API to connect with every provider is extremely powerful because it greatly reduces developer effort in building each integration individually. However, the increase in developer efficiency comes with the tradeoff of coverage.

Unifying all APPs within a SaaS category is hard work. As a Unified API vendor, you need to understand the APIs of each APP, translate the various fields available within each APP into a common schema, and then build a connector which can be added into the platform catalogue. At times, unification is not even possible, because APIs for some use cases are not available in all APPs.

This directly leads to low API coverage. For example, while Hubspot exposes a total of 400+ APIs, the oldest and most well-funded Unified API provider today offers a Unified CRM API which covers only 20 of them, inherently limiting its usefulness to a subset of the possible integration use cases.

Coverage is added based on frequency of customer demand and as a stop gap workaround, all Unified API platforms offer a ‘passthrough’ feature, which allows working with the native APIs of the source APP directly when it is not covered in the Unified model. This essentially dilutes the Unified promise as developers are required to learn the source APIs to build the connector and then maintain it anyways, leading to a hit on developer productivity.

So, when you are evaluating any Unified API provider, beyond the first conversation, do dig deep into whether or not they cover for the APIs you will need for your use case.

If they don’t, your alternative is to either use the pass throughs, or work with embedded iPaaS tools - both can give you added coverage, but they tradeoff coverage with developer efficiency, as we will learn below.

The Developer Efficiency challenge with embedded iPaaS

While Unified APIs optimise for developer efficiency by offering standard 1: many APIs, embedded iPaaS tools optimise for coverage.

They offer almost all the native APIs available in source systems on their platforms for developers to build their integrations, without a unification layer. This means developers looking to build integrations on top of embedded iPaaS tools need to build a new integration for each new tool their customers could be using. Not only this requires developers to spend a lot of time in API discovery for their specific use case, but also then maintain the integration on the platform.

Perhaps this is the reason why embedded iPaaS tools are best suited for integrations which require bespoke customization for each new customer. In such scenarios, the value is not in reusing the integration across customers, but rather the ability to quickly customise the integration business logic for each new customer. And embedded iPaaS tools deliver on this promise by offering drag drop, no code integration logic builders - which in our opinion drive the most value for the users of these platforms.

**Do note, that integration logic customization is a bit different from the ability to handle customised end systems, where the data fields could be different and non-standard for different installations of the same APP. Custom fields are handled well even in Unified API platforms.

What sets Knit apart?

So, we now know that the two most prominent approaches to scale product integrations today, even though powerful for some scenarios, might not be the best overall solutions for your integration needs.

However, till recently, there didn’t seem to be a solution for these challenges. That changed with the rapid rise and availability of Generative AI. The ability of Gen AI technology to read and make sense of unstructured data, allowed us to build the first integration agent in the market, which can read and analyse API documentation, understand it, and orchestrate API calls to create unified connectors tailored for each developer's use case.

This not only gives developers access to 100% of the source APPs APIs but also requires negligible developer effort in API discovery since the agent discovers the right APIs on the developer's behalf.

What’s more, another advantage it gives us is that we are now able to add any missing APP in our pre-built catalogue in 2 days on request, as long as we have access to the API documentation. Most platforms take anywhere from 2-6 weeks for this, and ‘put it on the roadmap’ while your customers wait. We know that’s frustrating.

So, with Knit, you get a platform that is flexible enough to cover for any integration use case you want to build, yet doesn’t require the developer bandwidth required by embedded iPaaS tools in building and maintaining separate integrations for each APP.

This continues and builds upon our history (however small) of being pioneers in the integration space, right since inception.

We were the first to launch a 'no data storage' Unified API, which set new standards for data security and forced competition to catch up — and now, we’re the first to launch an AI agent for integrations. We know others will follow, like they did for the no caching architecture, but that’s a win for the whole industry. And by then, we’re sure to be pioneering the next step jump in this space.

It is our mission to make integrations simple for all.

‍