Freshservice Freddy AI in 2026

The Freddy AI Agent allocation of 1,200 sessions per Enterprise license per year is frequently misunderstood. The 1,200 figure is per Enterprise license, not per end-user, and the sessions pool across every Enterprise license on the account. A "session" is any one unique end-user's interactions with the AI Agent within a 24-hour window, so a user who contacts Freddy three times in one day uses one session, not three.

For internal IT service desk use cases (the primary audience for this reference), the pooled allocation is usually generous. An organisation with 25 Enterprise agent licenses has 30,000 pooled sessions per year. A deployment serving 1,000 employees with 50,000 annual tickets and 40% deflection draws roughly 20,000 sessions, comfortably inside that pool. Because the pool scales with the number of agent licenses you buy, the allocation rarely binds for internal IT.

The allocation becomes relevant for CX-heavy environments (external customer support with high interaction frequency) or lean teams running few agent licenses against very high end-user demand. If your session demand approaches the pooled allocation, additional sessions are sold in packs (contact your Freshworks account manager for overage pricing). Model the session arithmetic before committing if you run a small agent team against a large user base.

Freddy AI Copilot (agent-facing, Pro+ or $29 add-on): suggested ticket responses, article recommendations, ticket summarisation, conversation sentiment analysis, and resolution note drafting. Copilot does not autonomously act; it assists the human agent.

Freddy AI Agent (end-user-facing, Enterprise-only): conversational self-service via Slack, Teams, portal, and mobile. Handles password reset, access requests, software installation requests, status queries, and KB article surfacing. Action execution (actually resetting a password, provisioning access) requires Freshservice Workflow Automator configuration with your identity provider, adding implementation complexity.

Freddy AI Insights (Enterprise-only): analytics on deflection rate, ticket trend detection, agent-performance analysis, and knowledge-gap identification. Useful for reporting to stakeholders but not load-bearing for the AI capability itself.

Freshservice Freddy AI deploys in 2-8 weeks on an existing Freshservice instance. This is the fastest deployment timeline in the enterprise AI ITSM category by a significant margin. The baseline setup (enabling Copilot for agents and the AI Agent virtual endpoint for users) takes days, not weeks. Configuration of knowledge base connectors, intent scope, and channel activation (Slack/Teams) typically adds 1-3 weeks. Workflow Automator integration for action execution (the part that moves Freddy from answering to doing) adds 2-4 additional weeks.

Freshservice's fast deployment is possible because the platform tightly integrates the AI layer with the core ITSM ticketing system. There is no complex middleware to configure. The constraint is KB readiness: if your Freshservice knowledge base is sparse or out of date, Freddy will hallucinate answers or escalate everything. KB readiness is the implementation task that determines whether a 2-week or 8-week timeline applies.