Paris, 05 February 2026 – Sopra Steria announces the launch of agentic workflow capabilities within its IAKA artificial intelligence platform. This major evolution now makes it possible to orchestrate AI agents capable of chaining multiple analysis and production steps within complex business processes, fully integrated into existing systems and under operator control.
Designed for critical environments — security, defence, justice, energy, transport and public services — IAKA thus shifts from an intelligent assistance model to coordinated execution of high‑value operational tasks, within a sovereign framework.
A new generation of automation: supervised AI action chains
The agentic workflow enables the coordination of several specialised agents capable of processing multimodal data, cross‑checking information, proposing intermediate decisions, and contributing to the production of structured deliverables.
These agents do not operate in isolation: they are orchestrated in complete processing sequences integrated into business processes. Each step is traceable, explainable, and verifiable, ensuring operational control, security, and compliance.
The platform offers ready‑to‑use agentic scenarios, adaptable to customer contexts, to accelerate the handling of complex situations and reduce manual processing workloads.
“Critical sectors need AI that acts — but above all, AI they can rely on with full confidence. By introducing agentic workflows into IAKA, its AI dedicated to critical sectors, Sopra Steria is paving the way for responsible automation of the most sensitive processes, combining technological power, human supervision and sovereignty requirements,” says Thierry Lempereur, Defence & Security Director at Sopra Steria.
From complaint to report: the judicial workflow use case
Presented at the Milipol exhibition, a concrete use case demonstrates the impact of agentic workflows on the processing of a complete judicial procedure. The agentic chain covers the entire journey, from the initial complaint to the production of the investigation report:
- Automatic, structured transcription of an oral exchange with a complainant
- Translation and normalisation of content if required
- Drafting suggestions for a compliant complaint statement
- Analysis of images and visual evidence associated with the case
- Automatic description and categorisation of collected elements
- Search and cross‑referencing in existing documentary databases
- Structuring of the procedure and tracking of key elements
- Generation of a synthetic, actionable investigation report
At each key stage, the human operator retains validation and decision‑making authority. AI handles transcription, structuring, analysis and document consolidation, delivering significant gains in time, reliability and completeness, while improving the overall quality of case files.
An agentic AI integrated into business systems and operated in a sovereign environment
IAKA stands out through its direct integration with business information systems and its ability to orchestrate AI processing within full operational workflows, far beyond simple conversational uses.
The platform is developed, hosted and operated in France. It meets the highest requirements in terms of sovereignty, security, traceability and governance expected in sensitive environments. This approach enables organisations to deploy agentic AI capabilities in a controlled manner, with strong alignment to their operational and regulatory constraints.
“This represents a strategic milestone for operational AI at Sopra Steria. With the agentic workflow, IAKA now enables generative AI to power complete business task sequences. We move from AI that assists to AI that contributes to producing, structuring and securing operational deliverables — always under human supervision,” says Guillaume Blondel, AI Expert Leader France at Sopra Steria.
With this evolution, Sopra Steria strengthens its position as a leading provider of sovereign AI for critical environments, focused on concrete, integrated and directly deployable use cases.