Genesys is partnering with AI lab Scaled Cognition to develop reliable agentic AI. The collaboration combines Genesys Cloud with specialized large action models to eliminate hallucinations and improve governance.
The collaboration focuses on solving a core problem with AI agents: unreliability. Scaled Cognition develops large action models (LAMs) that go beyond traditional chatbots. The company’s Agentic Pretrained Transformer (APT-1) leverages new technology to eliminate hallucinations.
“Trust, reliability and performance are essential for organizations to embrace agentic AI,” said Olivier Jouve, chief product officer at Genesys. The combination with Scaled Cognition should help organizations deploy AI agents without the risks currently associated with them.
The technology differs from ordinary large language models. Whereas LLMs focus primarily on text prediction, Scaled Cognition develops models specifically for decision-making and executing actions in customer experience environments.
The new platform should help organizations convert complex business actions into predictable behaviors. This keeps customer outcomes transparent and consistent with business standards.
Earlier this year, Genesys Cloud AI Studio and AI Guides were released. These tools enable organizations to set up semi-autonomous AI agents without programming knowledge.
Tip: AI agents take control of customer experience at Genesys
Strategic investment
The collaboration will initially focus on customer service, where AI agents can perform autonomous tasks while meeting compliance requirements. Ultimately, this should lead to fully autonomous agents working alongside both AI and human teams.
For organizations, this means that they will soon be able to deploy AI agents that provide answers and take action without human intervention. The reliability promised by Scaled Cognition could be the breakthrough that agentic AI needs to become widely accepted in business environments.