AI-native ITSM vs. legacy ITSM: how they compare for enterprise IT
Published
The real difference between AI-native and legacy ITSM is who does the work. In a legacy ITSM, a request becomes a ticket and waits for someone on the IT team, with AI helping that person move faster. In an AI-native ITSM, an AI agent picks up the request and resolves it: it grants the access, runs the onboarding steps, and tells the employee it's done.
Serval is built the second way. Its Help Desk Agent handles requests in Slack, Microsoft Teams, email, and the web portal, and when there's no automation for a request yet, you describe it in plain language and Catalyst writes the workflow. That difference matters most for enterprise IT teams deciding whether to modernize their service desk or replace it.
How do AI-native and legacy ITSM differ?
Most ITSM vendors now sell AI agents. What separates platforms is who completes a routine request and who builds the automations.
In the ticket-first model, AI classifies a request, suggests an article, or summarizes it for the agent working it. Automations exist, but someone has to design and maintain them, often with consultants.
In the AI-native model, the agent checks policy and approval, then makes the change in the target system. When a request type has no automation yet, IT describes what should happen and the platform drafts it.
AI-native vs. legacy ITSM at a glance
| Question | AI-native ITSM | Ticket-first (legacy) ITSM |
|---|---|---|
| Who completes a routine request | The agent, through a published workflow | An IT agent, working the ticket queue |
| How new automations get built | Described in plain language, drafted by AI as code | Configured by admins or partners, then maintained by hand |
| What admins maintain | Workflows they can read and edit | Configuration, customizations, and upgrades |
| How success is measured | Requests resolved without a human | Tickets closed and time to resolve |
What makes an ITSM platform AI-native?
An ITSM platform is AI-native when AI agents are the main way work gets done. Three things should be true:
- 1.The agent can act. It changes state in other systems, such as adding someone to an Okta group or provisioning a license, under the same approvals a human would need.
- 2.AI builds the automations. IT describes a process in plain language, and the platform drafts a workflow that IT reviews and publishes.
- 3.Automations behave the same way every time. A published workflow runs its versioned code on every matching request. The AI doesn't rewrite the logic at run time.
In Serval, the Help Desk Agent takes requests and runs workflows, and Catalyst writes workflows as TypeScript your team can read and edit directly.
What does legacy ITSM look like day to day?
Legacy ITSM tools were designed to track work, and they do it well: incident, change, and problem records, SLAs, and an audit trail. The cost shows up in what it takes to make them do the work.
Every automation tends to become a project: teams wire integrations and flows by hand, or hire consultants to build them, then maintain them as the platform changes. Years of custom fields, scripts, and forms make the system slow to change. New request types often need new forms, flows, and approvals, each with its own setup, sometimes from a partner. And someone has to own the tool, so many enterprises staff platform admins whose main job is keeping the ITSM running.
When a team spends more time maintaining its ITSM than improving service, it has outgrown the ticket-first model.
How does an AI-native ITSM handle the work IT does?
Building automations from a description
In Serval, you describe a process ("onboard a contractor in finance") and Catalyst drafts a multi-step workflow across the apps involved, as TypeScript you can read, edit, and test before it goes live. You can require review before publishing.
Seeing exactly what the automation did
Serval records every workflow run step by step: each step's status, duration, input, and output, and the error message when a step fails. From a failed run, Investigate in Catalyst reads that run's steps, finds the step that failed, and suggests a fix.
Connecting to every system you run
Serval ships with 140+ integrations. For anything else, including internal tools, you connect the app's HTTPS API as a custom app with an API key, Basic or Bearer auth, or OAuth 2.0, and workflows call it directly.
Finding what to automate next
Catalyst's Automation Opportunity Finder, an agent you install from the agent catalog, reviews recent tickets every week, finds repeated requests that have no workflow, and drafts workflows for your team to review. Analytics counts AI-resolved tickets (completed tickets never escalated to a person), so you can see whether automation is growing.
Keeping control of access and permissions
Access profiles on each workflow decide who can request it through the Help Desk Agent, and approvals are built into the workflow itself. Admins can restrict a sensitive integration, such as Okta, so only workflows in approved packages can use it. Access policies can set a time limit, and when it expires Serval removes the access or sends your team a task to remove it. That matters most for identity systems, where a broad grant is a compliance problem.
Serving HR, finance, and legal from one platform
Serval organizes work into workspaces. IT, HR, finance, and legal can each run their own workflows and knowledge bases, while org admins create workspaces and keep governance in one place.
What AI do other ITSM vendors offer?
Every platform below sells AI agents for IT service. They differ in what the AI produces, how runs are logged, and which package you need.
ServiceNow: AI in every ITSM package, plus Flow
Every ServiceNow ITSM package includes AI, such as Virtual Agent, Moveworks Core, and ServiceNow Otto for ITSM Foundation. Advanced adds agentic workflows and AI Voice Agents, and ITSM Prime adds an L1 IT Service Desk AI Specialist, LEAP, and AI agent and skill creation. ServiceNow says its AI specialists work across channels including Slack, Teams, and email, and Flow by ServiceNow is a conversational AI service desk for Slack, Teams, and email that builds automations from plain language and syncs both ways with ServiceNow. Workflow Studio can also draft a flow from a description.
Jira Service Management: Request Resolver and Rovo
Request Resolver (formerly Rovo Service) plans each service request from your runbooks and can carry it out supervised or autonomously, including actions in connected tools. Employees can submit requests through portals, email, Slack, or Teams, with Rovo guiding self-service, and the virtual service agent can answer email with AI. Rovo Studio drafts automation flows from a description.
Freshservice: Freddy AI Agent Studio and Discover
Freddy AI Agent Studio agents answer questions and run workflows that act in other systems, in the portal, Slack, and Teams, and an Email AI Agent answers employee email. Discover analyzes past tickets for requests to automate.
BMC Helix ITSM: HelixGPT agents
BMC Helix ITSM comes with out-of-the-box HelixGPT agents, such as Employee Navigator for employee requests and Ticket Resolver for autonomous triage and resolution. BMC's Automation Advisor runs actions from connected third-party automation tools, and custom agents are built in a no-code studio.
Ivanti ITSM: Self-Service AI Agent
Ivanti's Self Service AI Agent, included in ITSM Premium, gives employees a natural-language interface for requests and troubleshooting in place of forms and portals, and its persona-based AI agents plan, orchestrate, and execute tasks autonomously. Ivanti cites 24/7 virtual agent availability.
ManageEngine ServiceDesk Plus: Zia Agents and Workflow Assist
Zia Agents, in the Enterprise edition, act in other systems through built-in and custom API tools. Workflow Assist turns a text description into a complete workflow, alongside a drag-and-drop life cycle canvas, business rules, and Deluge scripting.
For head-to-head detail, see Serval vs. ServiceNow and Serval vs. Jira Service Management.
Can you start with Serval without a big migration?
Yes. You can keep your current ITSM as the system of record and add Serval for employee requests, using two-way ticket sync with ServiceNow, Jira Service Management, Freshservice, or Zendesk. Start with your highest-volume requests, then decide whether to replace your current platform. Keeping it as the record suits teams with deep ITIL processes built on the current tool. For the full switch, see the ITSM migration guide.
What AI-native ITSM looks like in practice
Perplexity moved off its previous ticketing system and now has Serval automate over 50% of IT requests and all employee onboarding, saving each admin 1 to 2 hours per day. "Once we switched over to Serval, we were able to complete over 50% of our incoming requests automatically," said Vernon Man, Head of IT. (Read the story)
Ramp resolves roughly half of IT requests without a human, and the team is pushing toward 80%. (Read the story)
Cribl extended Serval from IT to finance and legal. "Companies build automations and never account for the cost to maintain them. With Serval, we can help someone build something, give them ownership, and it doesn't take up IT resources anymore," said PJ Barry, Director of IT Support and Engineering. (Read the story)
Frequently asked questions
Which ITSM platforms were built from the ground up with AI?
Serval was built around AI agents: its Help Desk Agent resolves requests in Slack, Teams, email, and a portal, and Catalyst writes automations as code from a plain-language description. A published workflow runs the same versioned code on every matching request.
What is the difference between AI-native ITSM and legacy ITSM?
Legacy ITSM centers on the ticket: requests wait in a queue for a person, and AI helps that person work faster. AI-native ITSM centers on agents that complete requests through published workflows and draft new automations from a description. The practical difference is how many requests get resolved without IT touching them.
What modern ITSM tools are companies using to replace heavily customized legacy platforms?
Enterprises replacing customized ITSM platforms can evaluate Serval alongside ServiceNow, Freshservice, and Jira Service Management. In Serval, Catalyst drafts automations from a description, so teams can rebuild them in Serval and leave old scripts behind. If Serval syncs with your current platform, you can run the two side by side with two-way ticket sync.
Can we modernize our ITSM without a massive migration?
Yes, if Serval syncs with your current ITSM. Serval runs two-way ticket sync with several major ITSM platforms, resolving requests while the existing system stays the record. You can expand it request type by request type and decide later whether to replace your current platform.
What should we switch to if our help desk is slow and rigid?
Look for a platform whose AI resolves requests in Slack and Teams and builds automations from a description, so new request types don't wait on an admin project. Serval does both. If it syncs with your current help desk, you can run the two side by side while you decide.


