How Five9 Made AI Knowledge Connectors Self-Serve with Paragon

About Five9

Five9 is a cloud contact center company whose AI-powered Genius CX Platform helps more than 4,000 organizations manage their customer experience.

Its generative AI product, Five9 AI Knowledge, turns a company’s own content into real-time answers during live customer conversations, powering both virtual agents and human agents. But AI Knowledge is only as useful as the content it can reach, and that content is spread across systems like Salesforce, Google Drive, and OneDrive. For customers to bring it all into one place, Five9 needed reliable connectors to each of those systems.

Building those connectors in-house would have meant owning the hardest parts of every integration: authenticating with each third-party system, managing customer credentials, and keeping content in sync as it changes at the source. Multiply that across every system Five9 customers use, and integrations become a product of their own.

Instead, Five9 built its AI Knowledge connectors on Paragon. Paragon owns third-party authentication and sync orchestration, so Five9 engineers can focus on what makes AI Knowledge different, including the underlying AI models, RAG stack, integration UI and the access-control layer. The result is a faster path to adding new connectors as customers need them.

The promise behind AI Knowledge

Five9 wanted customers to stop worrying about where their knowledge lives. They should be able to point AI Knowledge at their content and keep agents supplied with accurate, up-to-date answers.

Getting there was the hard part. Customers could upload content to AI Knowledge manually, but that didn’t help the many organizations whose knowledge lived in multiple systems. A generic integration option filled some of the gap, but adoption still took significant setup effort.

Why not build the connectors in-house

The Five9 team considered building the connectors on its own internal tooling. However, that path came with a long tail of work including troubleshooting and maintaining every workflow, managing version control, and repeating the same configuration for each new app.

The bigger cost was focus. Every hour spent building and maintaining a growing set of different OAuth flows, sync engines, and retry systems was an hour pulled away from improving how AI Knowledge delivers accurate, up-to-date answers to agents. Five9 wanted its developers to work on the core product, not rebuilding integration infrastructure app by app.

“Paragon let us offer a ‘connect once, stay in sync’ experience without turning our engineering team into an integrations team. We wanted the integration specialists to do that part of the job.”

Yang Lin, Lead Product Manager, Five9

Paragon gave Five9 the speed and reliability it was looking for. Instead of building each connector from scratch, the Five9 team starts from a pre-built template, with managed authentication and built-in retries already in place, so new connectors ship faster and hold up in production. Paragon also brought infrastructure-as-code, SOC 2 and GDPR compliance, and responsive developer support, a foundation Five9 would otherwise have had to build and maintain on its own.

Enterprise security, deployed globally

Paragon completed Five9 vendor security review, including compliance verification and the company’s strict internal security requirements. That same rigor carried into deployment. Data residency means each region needs its own isolated instance, and Paragon’s regional deployment model let Five9 stand up instances across all regions in parallel instead of engineering a separate integration platform for each region from scratch.

With Paragon, Five9 brought these integrations to customers worldwide on the timeline it needed. Paragon’s commitment to both Five9’s security requirements and on-time delivery meant the team could move fast without cutting corners.

“When our security team flagged things that needed adjustment, Paragon’s team worked through them with us. That collaborative posture is the reason we were able to ship to production across all regions on the timeline we needed.”

Yang Lin, Lead Product Manager, Five9

A native, self-serve experience built to scale

Five9 didn’t want connectors that felt like a third-party add-on. Because Paragon exposes its full capabilities through APIs, Five9 could build a setup experience that looks and feels entirely its own. The team designed a guided flow inside AI Knowledge where admins pick a folder, choose tags, and set how files sync, all in the same interface they already use to organize knowledge. There’s no separate tool to learn, and setup happens entirely within AI Knowledge.

Customers can now connect their knowledge systems to AI Knowledge on their own, without waiting on services or setup help. That means faster onboarding and less work to get AI Knowledge answering from the content customers already have.

What Five9 built with Paragon is the kind of integration experience every AI product team wants to offer. Customers connect integrations such as Google Drive, OneDrive, and Salesforce to AI Knowledge on their own, in a setup flow that feels completely native, and built for scale, so customers can sync the knowledge articles they need for more accurate answers.

Using LLMs and retrieval augmented generation, AI Knowledge turns a company’s own content into contextual answers for agents in real time. Underneath it all, Paragon runs the authentication, sync, and retry infrastructure. That frees Five9 engineers to keep shipping new connectors and stay focused on the AI that sets its product apart.

“The most meaningful takeaway for us is that Paragon lets each new connector ship against a pattern we already trust, rather than requiring us to build the underlying auth, sync, and retry infrastructure from scratch every time.”

Yang Lin, Lead Product Manager, Five9

Advice for other product managers

Evaluate the partnership, not just the platform. Plenty of integration vendors can point to a long connector list. What set Paragon apart was how their team showed up during the build: flexible and quick to respond to new requirements, even mid-build, and willing to learn how Five9 customers actually use AI Knowledge instead of pushing a one-size-fits-all playbook.

“It’s easy to find a vendor with a great pitch. It’s much rarer to find one where building with them day-to-day matches that pitch.”

Yang Lin, Lead Product Manager, Five9

Building integrations into a product that has to clear enterprise security reviews? Talk to our team about running Paragon self-hosted in your own environment.