real-time AI

Gimlet Labs and Cerebras Target 3,000 Tokens/Sec Inference

Gimlet Labs and Cerebras Target 3,000 Tokens/Sec Inference

Application Development, Cloud, Government Technology /

Gimlet Labs and Cerebras have announced a collaboration to deliver ultrafast AI inference through the Gimlet Cloud, targeting 3,000 tokens per second for agentic and real-time workloads. The partnership pairs Cerebras’ wafer-scale CS-4 chip with high-throughput GPUs in a disaggregated inference architecture. ECI Research analyzes what the deal means for enterprise and government developers building production AI systems.

Gimlet Labs and Cerebras Target 3,000 Tokens/Sec Inference Read More »

Modulate Velma API: Real-Time Voice Intelligence for Enterprise

Modulate Velma API: Real-Time Voice Intelligence for Enterprise

AI, Application Development, DevOps, Enterprise Software, Machine Learning /

Modulate has launched the Velma Enterprise API publicly, giving developers access to real-time voice intelligence built on its Ensemble Listening Model architecture. The platform analyzes raw conversational audio for emotion, intent, behavioral risk, and compliance signals rather than relying on transcription. ECI Research sees AI agent oversight and compliance monitoring as the fastest-growing use cases for this emerging infrastructure category.

Modulate Velma API: Real-Time Voice Intelligence for Enterprise Read More »