Setup guide

Freshservice chatbot setup — what your instance needs before you enable Freddy AI self-service

The Freshservice chatbot (Freddy AI) can significantly reduce ticket volume — but only if the configuration underneath it is ready. This guide covers the prerequisites, the pre-go-live checklist, and how to diagnose your readiness before switching it on.

Check readiness before go-live

The free AI readiness audit scores your knowledge base, categories, and service catalogue against Freddy's requirements — in under 2 minutes.

Run free audit

Prerequisites for the Freshservice chatbot

Knowledge base

Required

The chatbot deflects tickets by surfacing relevant KB articles. Without adequate coverage, it returns generic non-answers and users log tickets regardless.

  • At least 40 articles covering your most common ticket types
  • Articles written in user-facing language (not agent-only documentation)
  • All articles updated within the last 6 months
  • No significant duplication — one article per topic, not multiple near-identical versions
  • Articles organised into categories that match your service structure

Ticket categories

Required

Chatbot-raised tickets need to route to the correct team. Flat or inconsistent categories mean Freddy cannot triage accurately, and agents end up manually reassigning.

  • At least 3-level hierarchy: category → subcategory → item
  • Consistent naming conventions across all categories
  • No legacy categories unused in the last 90 days
  • Less than 5% of recent tickets left at top-level category only

Service catalogue

Recommended

If you want users to raise requests through the chatbot (not just find KB articles), service catalogue items are required. Without them, the chatbot can only point users to documentation.

  • At least 10 service items covering common requests (software access, hardware, account changes)
  • Each item has a clear description users can understand without ITSM knowledge
  • SLA policies linked to each service item
  • Approvals configured for items that require manager sign-off

SLA configuration

Recommended

Chatbot tickets need their own SLA treatment. Users who self-serve expect faster acknowledgement than traditional channels — a shared SLA with other channels will produce misleading breach metrics.

  • A specific SLA policy for chatbot-raised tickets (typically lower urgency than phone/email)
  • Response and resolution times that match the self-service channel expectation
  • Escalation rules on any chatbot ticket that remains unresolved after 2× the first response SLA

Pre-go-live checklist

Knowledge base has 40+ articles covering top ticket types
Articles are in user-facing language, updated within 6 months
3-level category structure configured and in use
Legacy unused categories removed or retired
Service catalogue covers top 10–15 recurring request types
SLA policies exist for chatbot-raised tickets
Freddy Self Service enabled in Admin → Channels
Chatbot tested with 10+ common queries before go-live
Deflection rate baseline recorded before go-live
Post-go-live review scheduled for 30 days after launch

Run the automated chatbot readiness check

Instead of manually working through this checklist, connect your Freshservice account and get a scored breakdown of every prerequisite in under 2 minutes. The audit reads your configuration data directly — no ticket content, no personal data, read-only access.

Knowledge base

Article count, coverage, freshness, duplication rate

Ticket categories

Hierarchy depth, naming consistency, unused categories

Service catalogue

Item count, SLA linkage, approval configuration

Run free chatbot readiness audit
What does the Freshservice chatbot need to work?+

The Freshservice chatbot (Freddy AI) needs three things to deliver value: a knowledge base with sufficient article coverage for your common ticket types, a service catalogue with items users can request through the chatbot, and ticket categories that allow Freddy to route chatbot-raised tickets accurately. Without these, the chatbot will give generic non-answers and ticket deflection rates will be very low.

How many knowledge base articles do I need before enabling the Freshservice chatbot?+

There is no hard minimum, but meaningful deflection typically requires at least 40–60 articles covering your most common ticket types. Articles should be structured with a clear title that matches the language users would type, and updated within the last 6 months. The chatbot surfaces articles by matching user query language against article titles and content — thin or stale articles result in poor matches.

How do I configure Freddy AI chatbot in Freshservice?+

Freddy AI chatbot is configured through the Freshservice Admin panel under IT Service Management → Channels → Freddy Self Service. Before enabling it, ensure your knowledge base has adequate coverage, your service catalogue includes the items you want users to request via the chatbot, and your ticket categories are structured to handle chatbot-raised tickets correctly. Run an AI readiness audit first to identify configuration gaps before go-live.

Why is my Freshservice chatbot not deflecting tickets?+

Low chatbot deflection rates are almost always caused by knowledge base gaps. The chatbot deflects by finding a relevant KB article and presenting it to the user. If the article does not exist, is outdated, or is written for agents rather than end users, deflection fails. Secondary causes include service catalogue gaps (users cannot request through the chatbot because the item does not exist) and poor category structure (chatbot-raised tickets route to the wrong queue).

Is your Freshservice instance ready for the chatbot?

The free AI readiness audit tells you in under 2 minutes — scored across knowledge base, categories, canned responses, SLAs, routing, and service catalogue.

Run your free audit

Free · Read-only · Under 2 minutes