Key takeaways
• AssetWorks went from signing with Mosaic to seeing tangible results in 30 days.
• What started as an agent-assist use case expanded into Self-Service, knowledge automation, and a bug triage pipeline — all built on a single Mosaic platform.
• Mosaic connected to AssetWorks’ case management system, Confluence, JIRA, and legacy documentation going back to 2005 — no data migration, no developer resources required.
• Using Mosaic, AssetWorks built an agent that captures structured bug information directly from customers and populates JIRA automatically, where another agent handles triage without any custom development.
• The path to Mosaic came through the Volaris ecosystem, giving AssetWorks confidence in a partner already stress-tested across the portfolio.
The challenge: A knowledge problem hiding inside a backlog
AssetWorks builds software for operationally complex businesses that manage assets like vehicles, buildings, and infrastructure. Its products carry features accumulated over decades, each serving distinct industries, use cases, and customer expectations. For a technical support team, that kind of product depth compounds with every release. Every update adds functionality, edge cases, and configuration nuances that agents are expected to carry in their heads.
The result was a significant support backlog and a brutal ramp for new hires. No matter how capable they were, it took months before a new agent could confidently handle the full range of technical questions customers asked. In the meantime, the team leaned on its veterans, escalating internally through Slack and informal channels rather than finding answers independently. Institutional knowledge became a queue of its own.
The knowledge existed. It lived in Confluence, in years of JIRA history, in PDFs going back to 2005. The problem was that none of it was accessible at the moment an agent needed it. This is the defining problem of technical support at established software companies: the answers are somewhere, and "somewhere" is the bottleneck.
Kicking off a partnership with confidence
As a Volaris portfolio company, AssetWorks skipped the cold vendor search and was referred to Mosaic AI by peers who'd already deployed it, entering the partnership with a confidence that a standard evaluation rarely produces.
That’s how the team was introduced to Mosaic AI. The connection didn’t come from a cold outreach or a procurement cycle. It came from within a network of companies already making AI work in similar environments. For AssetWorks, that meant going into the partnership with a level of confidence that a standard vendor evaluation rarely produces.
“Going to somebody that actually had the tools off the shelf really got us into production much faster.” — Greg Richards, General Manager, AssetWorks
The solution: Starting narrow and letting the platform expand
The initial goal was to connect to Zendesk and the documentation sprawl, Confluence, JIRA history, and legacy PDFs, so new agents could surface accurate answers without pulling a senior colleague out of their own queue.
Mosaic AI indexed and structured that knowledge through its Customer Context Model, then put it to work inside the agent workflow. And once it was connected, something became clear: what Mosaic AI could do for an agent looking up an answer was not structurally different from what it could do for a customer asking the same question. Every solved problem revealed the next one, and each new use case ran on the same foundation. No new implementation or data migration.
The results: Reducing support costs by 39%
AssetWorks didn’t deploy Mosaic AI to solve a single problem. From the moment the platform connected to their systems, each solved problem revealed the next one, and the same Mosaic AI foundation supported every step without requiring a new implementation.
Assist: Answers on demand for technical support agents
Mosaic AI’s first job was making institutional knowledge usable. With Assist connected to Zendesk, Confluence, JIRA, and legacy documentation, new support agents could surface accurate, cited answers in seconds, drawing on the same knowledge that previously required interrupting a 10- or 15-year veteran.
The validation test was simple: could long-tenured employees ask Mosaic AI the questions customers actually ask and get reliable answers back? They could. That bar mattered because clearing it opened the door for customers to ask directly.
Self-Service: Deflecting cases before they become cases
Once agents had validated answer quality internally, AssetWorks extended the same knowledge foundation to customers through Self-Service. Customers now get accurate, generative answers to technical questions without opening a case, and every deflected case is one that never lands in the backlog the team was working to shrink.
“If we can make information readily accessible to brand new customer support agents, we can also do self-help for customers.” — Greg Richards, General Manager, AssetWorks
Knowledge: Turning recurring questions into coverage
Assist did more than help agents answer questions. It surfaced patterns in what customers were asking, which exposed where documentation was missing entirely. Mosaic AI identifies recurring questions to identify those gaps, then drafts candidate articles for the team to review and publish. Instead of guessing what to document next, the team works from evidence of what customers actually need.
“Let’s use the AI to crawl all of our cases. If something has come up more than two or three times, we know that’s probably a pervasive question. Let’s have the AI build a knowledge article about it.” — Greg Richards
Turning support data into an engineering feedback loop
As the case backlog shrank, a different problem came into view: engineering was consistently receiving bug reports without enough structured detail to reproduce them. The common complaint—“we don’t get enough information to build a reusable test case”—was a well-known friction point across the business.
Using Mosaic AI’s Agent Builder, the same no-code foundation underneath every Mosaic AI module, AssetWorks configured an agent that collects structured diagnostic information directly from the customer, determines whether the problem is likely a bug, and creates a JIRA issue with the detail engineering needs. A second agent handles triage. The full loop from customer conversation to engineering queue runs without manual intervention.
“We actually, using the same tool from Mosaic built an agent that elicits the information directly from the customer on the problem they’re seeing, determines if it’s a bug, and then actually populates that into JIRA—which now gets picked up by another agent that does the triage. We didn’t have to build any of that ourselves.” — Greg Richards, General Manager, AssetWorks
30 days to production
The measure of success wasn't only what Mosaic AI could do. It was how fast. From signed PO to tangible results in production, the timeline was 30 days. Not a pilot, not a proof of concept. Production impact in a single month.
That speed came from not having to build any underlying infrastructure. Mosaic AI connected to the systems AssetWorks already ran, extended across four use cases on one foundation, and left the team with a platform that keeps expanding rather than a project that needs constant maintenance.
“From the time I signed the PO with the company until we were actually seeing tangible results — honestly, it was only about 30 days.” — Greg Richards, General Manager, AssetWorks
