Press & media

FOR IMMEDIATE RELEASE
Monday 5 October 2026 · London, United Kingdom

Sentient Machines launches Emotional API, making emotional intelligence available to any application

Emotion analysis, longitudinal emotional profiling, and deepfake voice detection are now unified in a single API — available from a free developer sandbox through to full white‑label enterprise licensing

Sentient Machines has today launched the Sentient Emotional API, opening its emotion analytics engine to any organisation building software that involves human conversation. The API is available now at emotional-api.com. The launch makes three capabilities available through one interface.

Emotion analysis detects emotions, behaviours and events in calls, chats, emails, voice notes and AI-agent conversations, analysing tone of voice and language together rather than reading words alone. Longitudinal Emotional Profiling builds a personal emotional benchmark for every customer, so organisations can identify the people who have moved away from their own normal rather than from a population average. Voice Authenticity identifies synthetic and cloned voices, examining whether emotional reactions are consistent with genuine human behaviour.

Unlike conversation analytics platforms that require organisations to adopt a new dashboard, the Emotional API is designed to be embedded in systems companies already run — CRM, contact-centre platforms, business intelligence tools, mobile and web applications, and AI agents. It is delivered as a REST API with real-time webhooks, short-lived client tokens for consumer apps, and drop-in tool definitions for large language model agents.

“Every organisation is sitting on an untapped potential hidden inside the emotional history of its customers, recorded and stored, and almost none of it is used today,” said Danica Damljanovic, the Founder and CTO at Sentient Machines. “We’ve spent years teaching machines to hear what people actually mean. Opening that up as an API means a two-person start-up and a global bank can both build with it — and the person on the phone finally gets the context they need before the call connects.”

Emotional context before the conversation starts

Pre‑call context is expected to see the quickest adoption: one request delivers a customer’s emotional benchmark, current state, recent interactions and a recommended agent action before the call begins. Sentient Machines argues this changes contact‑centre economics, revealing emotional shifts that precede repeat calls and complaints — signals organisations can hear but currently cannot use.

Designed for multilingual operations

The API analyses conversations in the language they take place in — 99 languages across voice and text — and then returns transcripts and labels in the customer’s reporting language. The company says the order matters: translating before analysing flattens the sarcasm, hedging and distress markers that carry emotional meaning.

Responding to synthetic voice fraud

Voice cloning is now cheap, readily available and highly convincing, enabling account takeover, payment diversion and executive impersonation. Sentient’s method combines acoustic signatures of synthetic speech with emotional‑congruence analysis — checking whether a speaker’s reactions match what has just been said. For high‑risk moments, an active protocol called the Kerr Test issues short, natural prompts and measures the response. The company stresses that its outputs are risk indicators to trigger further verification, not proof of identity, with additional measures being actively developed to increase the detection rate further beyond the current accuracy of over 95%.

Responsible use built into the terms

Sentient Machines has published an acceptable use policy alongside the launch. It prohibits using emotional signals to pressure people into purchases or decisions, to drive credit, insurance, pricing or employment decisions, or to present outputs as lie detection or diagnosis. Indicators relating to customer vulnerability may not be used for pricing, targeting or ranking.

The platform also offers jurisdiction-aware processing, reflecting rules such as the EU AI Act’s prohibition on inferring employees’ emotions from biometric data. Customers send pseudonymous references rather than personal details, personal data in transcripts is redacted by default, and audio is deleted after processing unless retention is requested.

Availability and pricing

Free sandbox with results available immediately at emotional-api.com
Usage-based pricing across four units: audio minutes, text units, profile calls and deepfake checks
Pay-as-you-go for developers and smaller organisations; volume agreements for enterprises
Wholesale and white-label terms for software platforms, AI companies and resellers

About Sentient Machines

Sentient Machines builds emotional AI for organisations whose customers talk, type or speak to an AI. The Sentient Emotional API analyses voice and text together to detect emotions, behaviours and events; builds a longitudinal emotional profile for every customer so teams can see who has drifted from their own normal; and identifies synthetic or cloned voices. It is delivered as a REST API with webhooks, client tokens, no-code connectors and tools for AI agents. Available from a free sandbox to enterprise and white-label terms. emotional-api.com