When NelsonHall looked at Capgemini Enterprise Automation Fabric (EAF) a year ago, the primary expectation of clients and Capgemini’s emphasis was on monitoring and relating the impact of missed IT KPIs and incidents to individual business KPIs.

While this remains a very important foundation, clients are now typically looking to move beyond keeping the lights on to adopting solutions and tools that will help them drive innovation. Increasingly, this means incorporating greater analytics capability to take advantage of the large amounts of data that are being generated.

Accordingly, GenAI is increasingly important to enhance the service agent experience as well as the user experience and to begin the journey to autonomous execution incorporating self-learning and avoiding human intervention. Enterprises also increasingly want to understand their core application footprint and identify resolutions that support their ESG goals.

Fabric Philosophy Remains Key

To recap, Capgemini EAF is a plug-and-play configurable solution that addresses enhanced service management, observability, automation, analytics, and sustainability.

It provides an integration layer that connects the client landscape of ITSM tools, application and infrastructure logs, and monitoring tool outputs with Capgemini’s Manage Service Platform based on ServiceNow, which provides uniform governance, uniform processing, and intelligent dispatching.

The components used are specific to each client, with key functionality including:

• The Intelligent Dispatching module, an AI-based artifact that inspects incoming tickets and predicts the appropriate automation solution
• Service Management, supporting areas such as assisted resolution and problem management
• Observability, involving the creation of a CMDB and incorporating a noise reduction solution and a health check solution that provides a single pane of glass view of all health checks.
  One example of zero-touch automation is a health check created for price mismatches between a central pricing database and individual store POS systems. Here, the health check solution can detect discrepancies and create incident tickets; it can then potentially call a bot, self-heal the issue, and close the incident.

In other instances where self-healing is unavailable, the health check solution supports assisted resolution, using AI to recommend the most appropriate knowledge object to an agent

• Analytics, across ticketing data, events data, knowledge objects, and automation performance data
• Sustainability, providing functionality in support of carbon accounting.

In infrastructure, autonomous automation can handle around 30-50% of incidents without human involvement, typically exceeding 50% in the service request space. For application-related incidents, the level of autonomous resolution is typically in the range of 10%-15%.

EAF includes infrastructure-related automation bots for health checks, service requests, remediation, and reporting across:

• Servers (305 bots used across client engagements)
• SAP (85 bots)
• Storage & backup (65 bots)
• Network (48 bots).

In addition, custom automations are produced as necessary to support specific client applications. SAP represents 15-20% of infrastructure managed under EAF.

Introducing BuddyBot, a GenAI Service Assistant

Capgemini aims to make EAF an AI-first platform by augmenting the existing platform with sustainability and GenAI solutions. Three major themes are:

• Augmenting its current intelligent automation capability with more autonomous workflows and moving to a more agentic AI architecture
• Increased emphasis on sustainability to provide clients with the capability to understand the carbon footprint of the infrastructure being managed by Capgemini. The first MVP reports on the carbon footprint of the infrastructure related to the scope of work. The next MVP to be released will identify hotspots and provide “what-if” scenarios for reducing the carbon footprint
• Using GenAI to take advantage of the data on the platform to provide more contextual and personalized insights.

EAF’s five focus use cases for GenAI are:

• Operations insights
• Service desk operation
• Knowledge management
• Problem management
• GenAI accelerators.

Operations insights involve bringing all the available data points together and enabling the user to query how the service is operating in natural language. Previously, much of the reporting was static, showing data in tabular or graphic form.

To support service desk operations, Capgemini has introduced BuddyBot, a GenAI service assistant. It can listen to incoming conversations, undertake user identification, understand the context of the issue, and order ticket creation. It can then interface with the knowledge repository to generate suggestions, which the agent can use to solve the issue. The agent also has the option to provide feedback, which is used to enhance the usefulness of the recommendations.

Capgemini is currently keeping the human in the loop, but in the future, the GenAI assistant will likely be enabled to trigger automations where appropriate. Capgemini has also incorporated DataBricks onto the platform to improve the efficient processing of the data items and incorporate DataBricks AI functionality, providing deeper contextual insight.

Extending Deployment Options to Sovereign & Dedicated Models

Another key development is extending the deployment options for the Capgemini EAF platform. In addition to the multi-tenant model deployed on Capgemini’s AWS public cloud, Capgemini has now introduced:

• A sovereign model specifically targeting EU clients, with the platform only being able to be operated by EU citizens. The first customer is going live in February 2025
• A dedicated model, where the platform is deployed in a dedicated instance of the client’s hybrid cloud.

IT process automation elements can be deployed on the cloud or at client edge locations, allowing a more end-to-end approach to automation while addressing client security requirements, including data compliance and regulatory requirements.

Improving Event Identification & Assignment for Luxury Retailer

Capgemini assisted a U.K.-based luxury retailer that was experiencing slow resolution of issues related to its digital order management system. This system was based on IBM Sterling, SAP HANA, and other third-party software, including a shipping and warehousing system.

Two potential factors that were having a significant adverse impact were:

• The use of siloed monitoring tools without single pane-of-glass functionality and with health checks being done manually with little automation
• Tickets passing through multiple groups before landing in the appropriate assignment group.

To address these factors, Capgemini deployed:

• A solution using a CMDB to correlate events, achieving close to 90% prediction of the issue
• Automated health checks, reducing the number of manual checks
• Smart Dispatcher to predict the appropriate assignment group. Assignment to the correct queue grew to ~85%, removing the need to scale up the service team for special events.

Similarly, Smart Dispatcher has been deployed for a North American logistics company, where tickets, often raised by managers and involving the CRM application, were failing to go to the appropriate assignment group. This reduced the processing time from typically 4-5 hours to several minutes.

The Bottom Line

EAF is key to Capgemini’s management of IT infrastructure and applications on behalf of its clients. The key to its future is its continuing progress towards an agentic AI architecture with increasing levels of zero-touch resolution and natural language “what-if” analysis to identify remaining hot spots.