Bb5a31f81ClaimLegal AI platform Eve built autonomous voice AI agents to place outbound phone calls to insurance companies and retrieve client recordsAgreeDisagree
Bb5a31f81ClaimThe SAG-AFTRA collective bargaining agreement distinguishes between digital replicas of identifiable humans and fully synthetic performersAgreeDisagree
Bb5a31f81ClaimIn one operational mode, Eline van der Velden acts as a live underlying performance pilot that translates her actions to Tilly Norwood on screenAgreeDisagree
Bb5a31f81ClaimSAG-AFTRA released an official statement asserting that Tilly Norwood is an uncompensated computer program trained on performer data rather than a human actorAgreeDisagree
Bb5a31f81ClaimUnder the SAG-AFTRA contract, producers agree not to use synthetic performers if a human being or a digital replica can perform the roleAgreeDisagree
Bb5a31f81ClaimParticle6 created the artificial intelligence performer named Tilly NorwoodAgreeDisagree
Bb5a31f81ClaimTilly Norwood's technical architecture integrates a large language model, custom neural audio generation, and real-time visual renderingAgreeDisagree
Bb5a31f81ClaimAline van der Velden's production company Particle 6 created the AI digital performer Tilly NorwoodAgreeDisagree
Bb5a31f81ClaimTilly Norwood is designed by creator Aline van der Velden as a 24-year-old British digital personaAgreeDisagree
Bb5a31f81ClaimAline van der Velden conducted approximately 2,000 generation iterations to create the visual and behavioral model of Tilly NorwoodAgreeDisagree
Bb5a31f81ClaimVoice models capable of handling complex, multi-threaded conversations with interruptions improved substantially with the release of Gemini LiveAgreeDisagree
Bb5a31f81ClaimGoogle released Gemini 3.8 Live, which includes live, cartoonish avatars to give the AI a faceAgreeDisagree
Bb5a31f81ClaimThe Gemini 3.8 Live platform allows organizations to create their own custom avatarsAgreeDisagree
Bb5a31f81ClaimVoice models are being trained to perform "latency-budgeted reasoning," where thinking time is capped to ensure a fast responseAgreeDisagree
Bb5a31f81ClaimThe requirements for consumer voice channels and enterprise voice channels are very different, with enterprises needing more regulation and auditabilityAgreeDisagree
Bb5a31f81ClaimA proposed model for enterprises is to use a third-party voice agent as a "front door" while owning the backend reasoning and brand intelligenceAgreeDisagree
Bb5a31f81ClaimBenchmarking voice agents is difficult because it requires evaluating multiple dimensions simultaneously, including latency, response quality, and reasoning abilityAgreeDisagree
Bb5a31f81ClaimMany publicly available voice agent benchmarks are based on synthetic data because real voice data is private and difficult to releaseAgreeDisagree
Bb5a31f81ClaimA realistic voice agent benchmark must include latency as a critical evaluation metricAgreeDisagree
Bb5a31f81ClaimBuilding voice agents for enterprises is more complex than for consumer use cases due to additional governance and requirementsAgreeDisagree