Sopra Steria is the latest vendor to announce a significant investment in GenAI/LLMs and its incorporation into its software engineering platform (Ingine).

We recently met with Mohammed Sijelmassi, who holds two roles at Sopra Steria: CTO and Digital Transformation Office. We discussed how Sopra Steria is deploying internally and helping clients externally with GenAI and LLMs.

There is undoubtedly a sense of urgency: Steria acknowledges that GenAI will likely disrupt the IT services industry within two years, and its clients’ industries. The company has seen its delivery teams using GenAI and wants to accelerate this initial movement at the group level. Accordingly, Sijelmassi is sponsoring an AI use case program around three priorities: application services and its software products; educating clients on using AI; and internally for sales and support functions.

Immediate Priorities include Level 1 Support and Software Engineering

Sopra Steria is assessing LLM use cases in application services. An immediate opportunity is level 1 application support, with LLMs complementing other tools, such as self-service portals and chatbots, to deflect inbound calls to the service desk. The intention is to use GenAI together with chatbots to bring higher-quality responses.

Other immediate opportunities are related to software development and reverse engineering, e.g., for creating software documentation from existing applications, reading through lines of code to understand what an application does, annotating lines of code with comments, or identifying the application’s requirements (with the intent of generating test cases).

Sopra Steria is also assessing use cases in software development; next year, following several months of preparation, the company will train LLMs on its code repositories, including its coding and architecture best practices and standards.

Governance Issues

GenAI will bring governance issues. The increased adoption of AWS CodeWhisperer, GitHub Copilot, Duet AI for Google Cloud, and Hugging Face will increase AI-generated code proliferation. This raises challenges such as:

• The proliferation of code will impact its readability and degrade the application performance, and potentially impact the CPU and memory of a mobile device
• IP issues. Sopra Steria wants to avoid code and business logic developed specifically for a client being reused by a different organization. This means controlling the training data, separating code and business rules owned by clients, and ensuring best development practices are embedded in the training data.

LLMs have implications beyond governance and will impact junior roles such as level 1 application support. The company will increase its new graduates’ upskilling investment and migrate them quickly to level 2 roles. GenAI might have an impact on the age pyramid.

Sopra Steria is embedding LLMs into its software engineering platform Ingine to spread GenAI usage rapidly. Ingine also includes enabling tools and services such as a development environment, DevOps, security, and testing tools accessible through an internal marketplace. Sopra Steria plans to have Ingine GenAI-enabled by the end of 2023.

Mid-Term Initiatives

Beyond software development, reverse engineering, and support, Sopra Steria is looking at mid-term application services initiatives, such as:

• Code migration, e.g., from COBOL to .NET or Java. This will come later, as programming language migration requires transformation beyond code and includes reengineering the application based on different architectural principles. Sopra Steria is exploring how to use reverse engineering to create software specifications and automatically generate new code
• QA/testing.

Sopra Steria is addressing increasing client demand for workshops to explore current GenAI use cases and understand the IT implications around application architectures through its consulting arm, Sopra Steria Next. The company is working with its clients on the EU’s AI Act. It believes that demystifying fears or fantasies around GenAI will take another year for its clients. Sopra Steria Next will help here.

The company is also deploying LLMs for several of its software products. For instance, it is currently combining GenAI and ML for its HR software products, such as identifying changes in payroll and HR regulation, reviewing its software products to look for defects, and improving quality.

Productivity Gains Ease the Talent Shortage

Sopra Steria envisions the coexistence of large generic and slower-performance LLMs with broad capabilities, and more focused ones specialized in specific topics and use cases requiring low latency. Examples are use cases where real-time applications, such as manufacturing or defense applications, favor speed rather than breadth of knowledge.

There is no doubt that GenAI will bring productivity gains to the IT services industry while presenting it with enormous challenges, not the least of which is demanding significant investments in training. Sopra Steria’s stance is that GenAI will ease the pressure on the shortage of IT talent.