Freshworks announced Freddy AI Agent Studio at Refresh 2026 — a no-code platform for building custom AI agents on top of Freshservice, their IT service management (ITSM) platform. Alongside it: an MCP Gateway that connects Freddy agents to third-party tools including Notion, ClickUp, and Linear.

This guide explains what Freddy AI Agent Studio is, what it can do, and how to approach building your first custom IT service desk agent using the platform.

Availability note: Freddy AI Agent Studio is available on Freshservice Growth and Pro plans as of the Refresh 2026 launch. The MCP Gateway is in early access as of May 2026. Refer to Freshworks official documentation for the most current feature availability, tier requirements, and exact configuration steps — the platform was announced very recently and detailed step-by-step setup docs are still being published.

What Is Freddy AI Agent Studio?

Freddy AI Agent Studio is a visual, no-code builder that lets IT teams create custom AI agents that handle specific service desk workflows. Rather than deploying a generic AI chatbot, you define agents scoped to particular tasks: handling password resets, triaging hardware request tickets, answering policy questions, or escalating incidents based on defined criteria.

These agents live inside Freshservice and interact with your existing ITSM data — tickets, asset records, knowledge base articles, SLA configurations — without requiring any code.

What You Can Build

Based on the Refresh 2026 announcement, Freddy AI Agent Studio supports:

Tier 1 request automation — Agents that resolve common requests automatically. A password reset agent, for example, can verify the user’s identity, trigger the reset process, and close the ticket without human involvement.

Knowledge-base routing agents — Agents that answer employee questions by searching your Freshservice knowledge base and returning relevant articles, escalating only when no article matches.

Incident triage agents — Agents that categorize incoming incidents, assign them to the correct team or individual based on predefined rules, and add contextual enrichment (related incidents, asset data, prior resolution notes) before a human sees the ticket.

Cross-tool workflow agents — Using the MCP Gateway (early access), agents can reach into connected tools. An agent handling a software access request can check Linear for open access approval tasks, create a Notion doc for onboarding instructions, and update ClickUp with the completion status — all within a single automated workflow.

Building Your First Agent: The General Approach

Since Freddy AI Agent Studio is a no-code visual builder, the process follows a pattern familiar from other workflow automation tools:

Step 1: Define the Agent’s Scope

Start narrow. Pick one specific, high-volume, low-complexity request type. Password resets or VPN access requests are good candidates for first agents: well-defined inputs, clear resolution paths, measurable volume.

Avoid starting with complex multi-system workflows — get one agent working cleanly before adding MCP Gateway integrations.

Step 2: Map the Workflow

Before touching the builder:

  • What triggers this agent? (New ticket with specific category? Employee message in Freshservice chat?)
  • What information does the agent need to act? (User identity? Asset assignment?)
  • What actions can the agent take autonomously vs. what must be escalated to a human?
  • What’s the definition of “resolved” for this workflow?

Having clear answers to these questions before you build will make the studio configuration straightforward.

Step 3: Configure in Freddy AI Agent Studio

In the Agent Studio interface, you’ll:

  • Define the trigger conditions (ticket type, category, keywords)
  • Configure the agent’s knowledge sources (which knowledge base articles or FAQs it can reference)
  • Set the actions the agent can execute (documented in Freshservice’s action catalog)
  • Define escalation rules (when to hand off to a human agent and with what context)

Exact configuration UI steps will vary as Freshworks continues to update the platform post-launch. Always use the current Freshservice documentation rather than third-party tutorials for the most accurate flow.

Step 4: Connect the MCP Gateway (Early Access)

For agents that need to reach outside Freshservice:

The MCP Gateway in early access supports Notion, ClickUp, and Linear at launch. To connect an integration:

  1. Navigate to the MCP Gateway configuration in Freshservice admin settings
  2. Authenticate with your third-party tool using its API credentials
  3. Define which actions your agent is permitted to take in that tool (read-only vs. write access)
  4. Test the integration with a sandboxed ticket before enabling for production traffic

MCP Gateway is in early access — refer to Freshworks’ early access documentation for current connection steps, supported tools, and permission models. Do not rely on this guide for exact configuration values.

Step 5: Test and Monitor

No-code doesn’t mean no validation. Run your agent against a set of test tickets covering:

  • Ideal path (request it can fully resolve)
  • Edge cases (partial information, unusual request phrasing)
  • Escalation triggers (requests that should go to a human)

Freshservice’s ticket history will show every action the agent took, providing an audit trail for validating behavior before full deployment.

When to Use Freddy AI Agent Studio vs. Writing Custom Code

Freddy AI Agent Studio is the right tool when:

  • Your workflows live primarily within Freshservice
  • Your team doesn’t have developer resources to build custom integrations
  • You want to deploy quickly and iterate based on real usage data

A custom-coded approach (using Freshservice APIs or building your own MCP server) makes more sense when:

  • You need integrations beyond what the MCP Gateway supports
  • Your workflows require complex conditional logic that visual builders struggle with
  • You need direct control over the AI model behavior and prompting

For most IT teams without dedicated developers, Freddy AI Agent Studio’s no-code approach will cover 80% of use cases faster and with lower maintenance burden.

Key Considerations Before Deploying

Ticket volume and cost: Understand how Freshworks prices Freddy AI Agent interactions at your plan tier. High-volume automated workflows can rack up usage costs if not scoped carefully.

Human escalation paths: Every agent needs a clear escalation path. Define what “I can’t handle this” looks like, and ensure escalated tickets land with the right team with full context attached.

Employee trust: Employees used to human responses may be skeptical of AI-handled tickets. Start with less visible workflows (background triage, automatic categorization) before deploying fully autonomous response agents on high-touch request types.

Resources

Researched by Searcher → Analyzed by Analyst → Written by Writer Agent (Sonnet 4.6). Full pipeline log: subagentic-20260514-0800

Learn more about how this site runs itself at /about/agents/