RAICC

Responsible AI Call Center

 

Description:

The project is a 36‑month initiative called Responsible AI Call center (RAICC), led by DareData, with the goal of developing responsible AI products for call center operations and validating them in NOS’s real operational environment. Its core idea is not just to automate interactions with LLM-based agents, but to do so with strong guarantees around privacy, compliance, explainability, robustness, and operational control.

RAICC is a consortium of 12 entities, including companies and research organizations. The project is positioned as applied R&D that should produce deployable systems, not only research prototypes, and it targets operational maturity up to TRL 8/9 through staged validation in the NOS call center.

The project plans four main products: Gen-OS, AVD.AI, PRA.SaaS, and NetSpeak. Gen-OS is the central platform, designed as an operational layer for generative AI agents with supervision, observability, incident handling, knowledge management, and human-in-the-loop intervention in critical flows. The other products cover responsible AI development support, privacy-risk assessment for call transcripts, and natural-language interaction with routers through specialized language models.

Scientifically, the project combines LLMs, SLMs, multi-agent systems, dialogue-flow discovery, narrative understanding, privacy-preserving processing, and responsible AI evaluation. A major emphasis is placed on improved model cards, synthetic data for testing and auditing, adversarial robustness, explainability, and structured methods for understanding call-center conversations.

 

Consult the project's Spec Sheet here:

English

 

Co-funded by:

For any additional information, please contact us using the inquiries form.