What the session covers
Recorded at DMWF Global 2025 (Digital Marketing World Forum) in London, this session sits on the “Emotional AI for Marketing: Understanding Consumer Sentiment” panel. The panel brief covered using AI to understand and predict consumer emotions, integrating emotional AI into personalised campaigns, sentiment analysis for loyalty and engagement, and the ethical questions emotional AI raises for consumer trust. Sheoli’s contribution brings the voice dimension: the same signals that let a brand detect frustration or delight in a call also constitute biometric and emotional data about a person, and voice forensics, the analysis of who is speaking and how, is now within reach of any team with an API key. Her argument is that emotional AI only earns trust when the data it reads is contained: masked before it reaches a third-party model, used for the purpose the customer expects, and logged so the organisation can show what it did. That is the same principle Konsole applies to every model call.
Summary written from the session brief and the published panel description; the recording itself has no captions.
What you’ll learn
- What emotional AI and voice analysis can and cannot infer from how a customer speaks and writes.
- Why emotion and voice signals are sensitive data in their own right, and how that changes consent and retention.
- Where consumer trust breaks: inference without disclosure, data reused for a second purpose, and emotional signals sent to models the brand does not control.
- The controls that make emotional AI defensible: minimisation, masking before the model, purpose limits and an audit trail.
Key takeaways
- Treat emotional and voice signals as personal data with the same care as a card number or a health record.
- Decide what a model is allowed to see before you decide what it is allowed to infer.
- Trust is a product feature: brands that can explain what they read, and show the log, keep the customer.
More sessions

Clip · 2 min
Runaway agents and why you need circuit breakers
A two-minute excerpt on how agentic AI changes the cost risk profile: loops, tool calls with no end state, and why autonomy needs boundaries.

Webinar · 50 min
Taming runaway AI spending: how to control LLM and agent costs
A 50-minute session on why AI bills spike, how token pricing really works, the five levers that bring spend under control, and why those levers belong in a gateway rather than inside each app.