How Moonnox shipped seven enterprise integrations in four weeks and built the data foundation for its AI Knowledge OS

Moonnox was built by former consultants who hit the same wall over and over: knowledge walking out the door when projects closed, teams delivering inconsistently despite real expertise, and time lost stitching together tools that almost worked.

The product they built captures and contextualizes that institutional knowledge across conversations, documents, delivery workflows, and project history, then turns it into collective intelligence. But it can only deliver on that promise if it can reach the data those teams already live in. That integration requirement forced a choice most platforms cannot navigate: a unified schema or user-level authentication. Moonnox needed both.

“Moonnox is an AI-powered operating system purpose-built for professional services and systems integration teams," said David Cotten, CTO for Moonnox. “We help firms capture, contextualize, and preserve the knowledge that typically lives across conversations, documents, delivery workflows, customer interactions, and project history, then turn that into collective intelligence."

The founding team built the product out of direct experience. CEO Robert Ong co-founded Moonnox alongside Cotten, who spent over a decade leading complex technology programs at PwC and Bluewolf before this.

Backed by M25, the team's premise is specific: when a project closes or a consultant moves on, the knowledge they built disappears, and most firms have no infrastructure to stop it. Moonnox is that infrastructure.

The challenge: Integrations are the foundation, not a feature

For Moonnox to deliver on that promise, it has to read from the systems its customers already live in: SharePoint for documents, Box for file storage, S3 for object data, Jira for project and ticket history, Confluence for knowledge bases, HubSpot for client relationships. If those connections fail or degrade, the AI has nothing to reason over.

"Moonnox is only as powerful as the context it can access and understand," Cotten said.

That meant integrations could not be a late-stage product feature or a series of one-off builds. They had to hold up at scale, stay secure, and stay maintained without consuming the engineering resources Moonnox needed to build its actual intelligence layer. 

That's what led Moonnox to Paragon.

"Paragon helps us connect to those systems in a scalable, secure, and maintainable way,” Cotten said, “so our engineering team can focus on building the Moonnox intelligence layer instead of spending years maintaining integration infrastructure.”

The stack that could not scale

Before Paragon, Moonnox ran on a combination of custom-built connectors, Merge, and other external providers. It worked, but not well enough.

"Before Paragon, we had a combination of custom-built integrations and other providers," Cotten said. "That worked to a point, but it wasn't the right long-term foundation for where Moonnox was headed."

The pain was both technical and economic. Every system Moonnox needed to read from came with its own API structure, permissions model, data format, and edge cases. Maintaining that surface area pulled engineering time away from the actual product, and the maintenance load only grows as the integration count does.

"The bigger issue was long-term maintainability,” Cotten said. “We didn't just need to connect to systems. We needed to normalize data, support secure authentication patterns, and reduce the ongoing burden of schema mapping and connector maintenance.”

The evaluation: Looking for the only platform that did both

Moonnox's customers are professional services firms, and the data they process is often tied to those firms' own clients. The evaluation criteria covered scale, security, flexibility, and long-term maintainability. Most platforms failed on at least one dimension.

"Paragon was the only tool we found that had a unified schema while still allowing for user-level authentication."

David Cotten, CTO, Moonnox

"Paragon was the only tool we found that had a unified schema while still allowing for user-level authentication. That combination was critical. It gave us a way to simplify ingestion and normalization while still meeting the security requirements we had for customer data access."

Most integration platforms force a choice between two things Moonnox needed together:

The catch: a normalized API layer usually flattens the authentication model along with the data model, so you get simplicity at the cost of security. Paragon's Managed Sync architecture keeps both: a unified schema for simplicity, and per-user credential scoping for security.

"Other tools could solve parts of the problem, but Paragon was the only one that checked all the boxes for us."

That combination is what Moonnox's architecture required.

A commitment written into the contract

Moonnox signed with two hard requirements that did not yet exist in the product. Paragon committed to specific ship dates for both in the contract.

"The fact that Paragon was willing to commit to a ship date gave us confidence that they were willing to be a real partner."

David Cotten, CTO, Moonnox

"The fact that Paragon was willing to commit to a ship date gave us confidence that they understood our requirements and were willing to be a real partner. This wasn't just about buying what existed on day one. It was about choosing a platform and team that could support where we were going."

The build: From schema mapping to real customer scenarios

Implementation focused on three things:

1. Core managed sync and authentication flows. Stand up the connection and credential layer first.

2. Schema validation. Confirm how Paragon's unified schema mapped into the Moonnox data model.

3. Real-world connector testing. Test connectors against actual customer scenarios, not just happy paths.

The edge cases appeared early. Moonnox's customers store documents across multiple SharePoint locations, nested project folders, and team-specific repositories. Per-user credentials added another layer: each end user needed their own authenticated connection, scoped to their actual access permissions.

"Multi-folder SharePoint sync was important for us because customer data doesn't always live in one neat location. Per-user credential scoping was also critical because of our security requirements. Paragon's support for user-level authentication, combined with the broader managed sync model, made that possible."

Both issues were resolved with the Paragon team working directly in Moonnox's Slack channel. That kind of embedded support is not a given during implementation.

The outcome: Putting it into production

Four weeks from signed contract to customers testing live. Seven integrations in production: Jira, SharePoint, Box, S3, OneDrive, Confluence, and HubSpot. The alternative would have looked very different.

“Building it in-house would have taken months, especially for the volume of connectors we are currently running," Cotten said. "It wouldn't just have been the initial connector development. We also would have needed to build and maintain authentication flows, schema mapping, sync infrastructure, permissions handling, monitoring, error handling, and ongoing connector updates. That maintenance burden was exactly what we were trying to avoid."

Integrations shift the sales conversation and establish trust

Getting integrations right changed how Moonnox talks to prospects.

"When we can tell customers that we have a reliable, secure way to connect to the systems they already use, the conversation shifts away from implementation complexity and toward business value," Cotten said.

For an AI product, that shift is not a minor sales advantage. It is the difference between spending a discovery call defending your architecture and spending it demonstrating what the product can do. Moonnox now leads with the outcome.

The bottom line: The true cost of infrastructure isn’t the build

If there's one thing Moonnox's experience makes clear, it's that the real cost of integration infrastructure isn't the build. It's everything that comes after.

"Be honest about the real cost of building integrations in-house," Cotten said. "The initial connector is only part of the work. The real cost is schema mapping, authentication, permissions, sync reliability, monitoring, edge cases, API changes, and long-term maintenance.

“If integrations are strategic to your product but not your core differentiator, it's worth finding a platform that lets your team move faster without taking on all of that infrastructure yourself. For us, Paragon gave us that foundation."

Building on enterprise data? See how Paragon's Managed Sync handles authentication, schema mapping, and sync infrastructure so your team doesn't have to.