Replicant is a Voice AI platform that allows machines to listen, think fast, and speak more naturally and intelligently than ever before. With a highly-skilled team of deep thinkers from technology and the arts, we’re tackling a big industry using powerful machine learning algorithms and a keen ear for the aesthetics of voice communication.
|2020||Sep 2020 05:31 PM||27.0||11660|
|2019||Jul 2019 06:01 PM||7.0||11660|
AI Technology Stack
Custom AI Models: Replicant’s proprietary transcription and intent classification engine are trained specifically on lossy 8khz customer support phone call data, yielding 94%+ classification accuracy – the highest in the industry. Customers can also train custom transcription models for specific use-cases with just a few hours of data, ensuring any industry expression or out of vocabulary words are accurately captured so that every conversation is tailored to your needs.
Multi-intents: A key part of Replicant’s conversation design capabilities is being able to resolve multiple intents in a single turn of a conversation. With Replicant, your callers aren’t limited to asking simple questions; instead, the next time a caller wants to change their billing address and inquire about a refund, they can – in a natural, conversational way.
Replicant uses active learning to intelligently and automatically create a data pipeline for retraining. Only the appropriate samples are escalated to human reviewers, increasing labeling efficiency. All model retrains are evaluated for regressions and provide full visibility into the performance of every new model and intent within it, ensuring that your conversation models are getting smarter every day.