Building universal search and integration-enabled AI employees with Paragon

Introduction

Frame.so is an AI-powered operating system for businesses that enables users to centralize their work and deploy AI Employees to automate tasks. We spoke to founders Nicolas Baranowski (CEO) and Maksim Kovalenko (Head of Engineering) to learn more about what they’re building, as well as their experience using Paragon.

"Our vision is to empower our customers to manage every aspect of their business through a unified collaboration suite, seamlessly integrated with connected AI employees,” - Nicolas Baranowski, CEO.

Since their launch, Frame has seen impressive growth, with 22,000 signups and a 22% month-over-month growth rate.

The Challenge

"Most of our customers’ business data, even in very small organizations, is spread across multiple applications. Prioritizing integrations allows them to start using Frame without the need to 'import' business data first." - Nicolas Baranowski, CEO

As Frame aimed to become the central nervous system for companies, so in order to go to market successfully, the team knew the needed to provide robust integrations to allow users to connect their external data seamlessly.

However, they wanted their engineers to be focused on the core product, which meant building these integrations from scratch was simply not feasible.

The Search for a Solution

Frame's engineering team, led by Maksim Kovalenko, began evaluating options for implementing integrations. They considered open-source solutions, building in-house, and various integration platforms.

"Initially, we started by looking at some open source solutions. The main issue with those tools, aside from just them being open source and you having to roll out the infrastructure and maintain all of that yourself, is they are mostly aimed at you automating your own environment. So they're not for multi-tenant systems."

The team's key requirements included:

1. A rich catalog of off-the-shelf integrations

2. Multi-tenant architecture support

3. Ease of implementation and maintenance

After careful consideration, Frame chose Paragon as their integration solution.

"Paragon offered a rich catalog of off-the-shelves integrations. The team was also very hands-on with the technical onboarding," Nicolas shares.

Implementation of Paragon

In terms of the actual functionality provided to their end-users, Frame is now able to offer these four key features through the integrations:

1. Cross-app search of core objects

2. Cross-app linking capabilities

3. Real-time awareness of external knowledge for AI employees

4. Specific document querying through the AI employees

To preserve the seamless end-user experience, they also used a headless implementation of Paragon, enabling them to fully control the configuration experience for their end-users.

Benefits and Results

Since adopting Paragon, Frame has already saved hundreds of hours of engineering on building and even more on debugging.

Time and Resource Savings

"So far we shipped five integrations by staffing one senior engineer per week per integration, which would have otherwise taken us weeks to do per integration, so a few hundreds of engineering hours already so far," Maksim notes.

This is possible because of the re-usability aspects of Paragon. "After you have one integration ready, you just basically copy paste the same steps and you have 60% of the work already done for you," Maksim points out.

Durable infrastructure

They’ve also found it helpful to offload all the execution-level challenges such as rate limits and error-handling. "Paragon handles retries, and some edge cases around that. This saves time because you may have users that have lots of documents that are connected, and you would otherwise need to figure out how to handle 3rd party rate limits.”

Streamlined Troubleshooting and Monitoring

"With Paragon, it takes at least 2x, if not 3x less time to figure out what an issue is," Maksim explains.

He contrasts how easy debugging is on Paragon with the experience building in-house. “If you’re developing yourself, the only thing you have are your server logs or things printed in the console. Then you have to figure out which user [the errors] belong to.”

Future Plans

Frame is actively working on expanding its integration offerings, as there are dozens of applications that their customers use.

These integrations lower the barrier to entry, and unlock the opportunity to eventually migrate Frame's users’ data into their own platform so Frame can become the all-in-one platform for work.

Alongside that, they are also working on providing the AI employees more granular permissions, as not all data should be accessible be everyone.

Nicolas shares their vision: "Our vision is that you'll be able to deploy any kind of AI employee, give them access to different pieces of software and data cubes, be it Jira or Gmail or DocuSign, and finally deploy commands for them to take action."

Conclusion

Frame's partnership with Paragon has been instrumental in realizing their vision of a unified, AI-powered business operating system. Integrations play a critical role in their product strategy, and Paragon has enabled them to significantly accelerate their roadmap.

Frame's implementation is a testament to how extensible Paragon's platform can be, and as they continue to build out their suite of tools and AI employees, we're excited to see what they do with the platform.