Challenge
Modo Networks needed a more consistent way to classify, prioritize, and enrich thousands of tickets each month. Technicians without deep tenure on legacy specialized systems had no efficient path to resolution for those tickets. Senior technicians were also spending too much time transferring institutional knowledge to Tier 1 instead of advancing higher-value project work.
Solution
Modular ConnectWise AI Agents™ were deployed within existing PSA workflows. The agents automate ticket classification, prioritization, and data enrichment at intake. Issue-specific agents were also built using technical documentation to extend senior-level expertise across the service desk. The rollout was deliberate and phased. Agents were activated one at a time, with each SOP validated against real ticket data before going live.
Results
CW AI Agents™ now handle the vast majority of ticket volume, with adoption reaching approximately 90% in recent months. The completed-to-closed QA cycle runs 30% faster and has materially fewer data errors. Automated priority-setting gives dispatch a cleaner context earlier. Technicians are also resolving complex issues that previously required escalation.
An operator’s approach to AI adoption
Modo Networks has operated as a Texas-based MSP since 2008. It has built a long-tenured technical team, including Chief Operating Officer Geoffrey Grafing, who has been with the company for more than a decade. Several technicians have advanced from entry-level roles to senior technical positions. That growth path shapes how Modo evaluates technology. The standard is whether a tool accelerates a technician’s development.
When generative AI reached the market, Modo’s technical staff and ownership, composed of practicing technicians, moved quickly to evaluate it against complex, unfamiliar documentation. The immediate value was comprehension: feeding dense vendor material into CW AI Agents and interrogating it for meaning closed knowledge gaps faster than any static knowledge base could.
The technicians who learn to use AI well are going to have an advantage. For us, the question became how to bring it into the company safely and responsibly, so our team could stay ahead.
Turning ticket intake into cleaner operational data
Modo first evaluated CW AI Agents with dispatch and assignment in mind. As the team tested the solution, its value expanded across the ticket lifecycle. CW AI Agents could standardize operational details across all incoming tickets. That gave dispatchers and technicians cleaner data before human assignment decisions were made. The agents now review incoming tickets across Modo’s boards and set the fields the business depends on: type, subtype, priority, SLA, company, contact, and configuration. They do this across approximately 6,000 tickets a month, with agents touching roughly 65% of total volume since deployment and as much as 90% in recent months.
We came in looking at dispatch, and what stood out was how much more the agents could do around the ticket. They take on the ticket details that make the service desk run cleaner—the fields, the priorities, the follow-through that technicians do not want to spend their time on.
Modo still treats the ticket assignment itself as a human judgment call. Leaders want dispatch decisions to stretch technicians and support their development. That gives CW AI Agents a clear role in the workflow. The agents automate the surrounding administrative layer, execute it consistently at scale, and provide people with better information for the judgment calls they still own.
Measurable gains in data quality and cycle time
Manual correction of ticket classification had been a recurring drain on Modo’s leadership team. CW AI Agents now accurately set type, subtype, configuration, and company data at intake. As a result, fewer errors reach the QA review stage. Tickets also move from completed to closed measurably faster, narrowing a gap where finished technical work had been waiting for administrative follow-through.
The ticket is cleaner by the time it gets to QA. We’re seeing far fewer errors, and that completed-to-closed category is 30% or more faster for us.
Automated priority-setting also improves the front end of the ticket lifecycle: because the agent assigns priority as soon as a ticket opens, dispatch has a clearer context sooner. Modo attributes gains to both ends of the ticket lifecycle. There’s less administrative overhead before technical work begins and less follow-up after it is complete.
Extending senior-level expertise to every tier
Modo’s original business case for AI focused on escalation load: senior technicians spending a disproportionate share of their time developing junior staff rather than advancing project work. The company addressed this directly by building issue-specific AI agents grounded in its own knowledge base. For example, an agent built around documentation for a large customer’s legacy Linux-hosted print infrastructure, which is a system requiring specialized knowledge that newer technicians had no efficient way to acquire.
On a legacy technology, a Tier 1 technician may not even know what symptom to look for. The agent gets them to the right pattern: here’s the problem, and here’s how you fix it.
Technicians report using the ConnectWise AI Assistant™ roughly 5–10 times a month, mainly for Tier 2 and Tier 3 issues. These are the moments when independent resolution is hardest. For Modo, the return shows up when a technician can close one of those tickets without pulling a senior technician away.
The value is not how many times someone opens the [ConnectWise] AI Assistant. It’s what happens on the more challenging tickets: when they use it, they’re able to close work that otherwise would have escalated.
The effect is evident on the leadership side as well. With classification handled reliably upstream, managers spend less time correcting ticket data. They can redirect that time toward higher-value technical mentorship for the smaller set of tickets that genuinely require it.
The advantage of AI is grounded in operational data
Modo’s technical staff draws a clear distinction between general-purpose AI tools and those natively integrated with the company’s environment, ticket history, and documentation. That distinction formed the basis for building an internal troubleshooting agent on the ConnectWise Platform™, trained on Modo’s operational data rather than public information alone.
The difference is our data in context. ConnectWise AI Agents understand our documentation and ticket history and can get to a root cause in a way general-purpose AI simply cannot.
A phased, evidence-based rollout
Modo’s guidance to MSP peers is practical: validate slowly and activate agents individually rather than enabling the full platform at once. CW AI Agents’ modular architecture made this possible. Because agents are discrete and task-specific, Modo could adopt AI within real service workflows and build confidence with each use case.
Being able to turn on one agent at a time made adoption practical. We could learn how it responded to our SOPs, improve the prompts, and build confidence before expanding.
Equally important was the ability to validate SOPs against real ticket data before production deployment. When an early routing rule proved too broad, Modo’s team reviewed the agent’s stated reasoning. The process was auditable and iterative.
When an agent makes a decision, we can see the rule it followed and how it applied it. That lets us revise the SOP, test it against the same ticket, and know the wording produces the decision we want.
Validating adoption empirically
To confirm genuine reliance on the platform rather than assumed usage, Grafing ran a controlled test by turning off select features to see whether technicians who reported minimal use would notice their absence.
Some technicians did not realize how much they were benefiting until I turned features off. Within days, they were asking for them back.
Grafing’s broader guidance for peer operators focuses on change management. Teams need consistent, transparent communication throughout the deployment. That helps AI be seen as a benefit the team understands, not as a change imposed on them.
Moving toward predictive, pattern-based service delivery
Modo’s leadership is explicit that AI’s role is to augment technician judgment, not to replace it. Grafing sees the strongest current value in using AI for repeatable operational details and the strongest future value in recognizing patterns across tickets before humans have to connect them manually.
The next step is pattern recognition. AI will be able to see three tickets that look unrelated to a person, flag the common issue, and give the technician that context before they start.
Modo is heading in this direction: using AI as the layer that keeps process knowledge up to date at scale, allowing technicians to focus their judgment on critical areas.
Key takeaways
- Start with the operational work that slows the service desk down: CW AI Agents can create value by handling the ticket details that technicians and managers spend too much time correcting.
- Use AI to improve data quality at intake: Consistent type, subtype, company, contact, configuration, priority, and SLA data give teams cleaner tickets and better downstream reporting.
- Keep human judgment where it matters: Modo kept assignment decisions with people while using CW AI Agents to strengthen the administrative layer around those decisions.
- Measure gains across the ticket lifecycle: Modo saw value at both ends of the process, from cleaner prioritization at intake to a faster completed-to-closed QA cycle.
- Bring senior knowledge closer to Tier 1: Issue-specific copilots help technicians resolve complex tickets by surfacing trusted internal knowledge at the moment of work.
- Roll out agents one workflow at a time: Activating agents gradually and testing SOPs against real tickets helped Modo build confidence before expanding.
- Use adoption signals beyond raw usage: The strongest proof was not how often technicians opened the AI copilot, but whether they missed it when it was gone and used it to close harder tickets.
