How Did CPage Reduce the Volume of Support Tickets by 20% Using AI?

The challenge: handling a growing volume of requests and relieving the burden on N1 support

Streamline and optimize business support

More than 15,000 annual requests for sophisticated and complex hospital software

As a Public Interest Group (GIP) and software publisher serving exclusively public healthcare facilities and Regional Hospital Groups (GHT), the CPage Group supports more than 320 members and 5,000 hospital users on a daily basis.

Business support handles a very high volume of requests:
The support team processes 15,000 tickets each year, 60% of which relate to business or application support. The HIS application solutions (i-GAP for the patient journey and billing, i-GEF for finance and procurement, and i-GRH for human resources) are both feature-rich and technically complex for non-expert users.

Strategic Challenges: Reducing the Volume of Tickets and Structuring Knowledge

Faced with this workload, CPage has set several key objectives to modernize its incident handling process:

• Relieve pressure on Support Team N1: Handle recurring, low-value-added questions independently to free up time for support experts to focus on complex cases.
• Ensuring the autonomy of hospital staff: providing an immediate, context-specific response within the software, without having to wait for a support ticket to be processed.
• Handle the increase in demand with the same number of employees: provide scalable support to keep pace with the publisher's growth.
• Leveraging knowledge assets: centralizing and utilizing information that is currently scattered across official documentation, internal procedures, and ticket history.

The goal: to rethink the incident handling process​ to reduce the volume of incoming tickets by 20%.

The solution: a sovereign AI chatbot integrated into ITSM

Position AI as the No. 1 and mandatory filter 

Optimizing Level 1 Support

To streamline the incident handling process and reduce the volume of support tickets, CPage has chosen to make AI the primary filter for support. This makes it mandatory to go through an intelligent chatbot before creating a ticket, while keeping the phone line as a safety net.

Integrated with the core of its ITSM platform (IWS by Isilog) and named EKKO, this intelligent chatbot is designed for several user groups:

• The primary target audience: users at participating hospitals, with the goal of providing them with immediate self-sufficiency in terms of business and application support.
• The secondary target audience: CPage support technicians, to make it easier for them to find information on a daily basis.
• Internal, opportunistic use: Although internal employees were not initially targeted, the platform was adopted naturally to facilitate information searches and speed upthe onboarding of new hires.

A secure, sovereign, no-code AI architecture

This chatbot was created and deployed using the Wikit Semantics platform, which is characterized by strict technical and organizational requirements:

• Sovereignty and Security: Platform hosted in France by OVHcloud and audited through external penetration tests.
• No-code & multi-model approach: simple administration with no technical skills required and access to the best LLMs/SLMs on the market.
• ITSM Connectors: Direct integration with the IWS knowledge base and history.

AI as a Cycle of Remediation and Continuous Improvement

To avoid errors, maintain a high level of trust, and optimize the chatbot’s responses, CPage relies on a four-step continuous improvement cycle:

1. Automated conversation analysis: systematic identification of queries for which the AI did not provide an optimal response.
2. Review of unsatisfactory responses: handling by support teams and direct follow-ups with affected users during the launch phase.
3. Source correction and enrichment: daily updates to the knowledge base.
4. Support handling: a smooth, simple, and traceable escalation process to a technician if human intervention is still required.

The Results: Tangible Results in Terms of Workload and Support Responsiveness

Faster, more accessible, and less overburdened support

Measurable key performance indicators (KPIs)

• 700 daily conversations handled by the EKKO chatbot
• -20% of incoming tickets starting in the second month of production, with discounts peaking at -30% to -35% in some months.
• A 33% reduction in the median resolution time, which has dropped from 1.5 days to just 1 day.
• Up to 40–50% of requests are resolved on their own, depending on the maturity of the knowledge base.

Uninterrupted service outside of business hours

20% of conversations take place outside of support hours (between 6 p.m. and 8 a.m.). The AI chatbot thus provides 24/7 first-level support for hospitals, where operations never stop.

Beyond the Numbers: Tangible Benefits for Users and Support Teams

• Time savings and greater autonomy for users: By enabling hospital staff to instantly find answers within the tool, this approach eliminates the wait time associated with processing a support ticket. The result: consistently high responsiveness across all inquiries, whether they involve technical issues or simple requests for assistance. :
• Comprehensive business support: Beyond its primary function of assisting with day-to-day inquiries, the chatbot fulfills two strategic roles:
  • Regulatory Issue Management: It provides users with a high degree of autonomy when dealing with complex or anxiety-inducing topics (e.g., the Electronic Invoicing Reform).
  • Major Incident Management: In the event of an application outage or slowdown, it immediately provides workarounds, thereby preventing the support system from becoming overwhelmed with duplicate tickets.
• A more rewarding role for technicians: Freed from repetitive and routine tasks, support teams can devote their time and expertise to complex incidents that require genuine human assistance.

CPage's Perspective

Key Factors for the Project's Success

CPage shares its recommendations

To ensure the success of this platform transformation project and maximize AI adoption, CPage has identified four fundamental pillars:

• Set measurable goals from the outset: define clear metrics during the scoping phase to concretely measure operational gains.
• Ensure an up-to-date and reliable knowledge base: maintain high-quality, constantly updated documentation to ensure the accuracy of the AI's responses.
• Ensure a comprehensive change management process: support both internal support teams and end users at the facilities (communication strategy).
• Adopt a continuous improvement approach: regularly analyze conversations to refine your understanding and improve the user experience over time.