
94: Voice AI & Conversation Design Explained: Dark Patterns, Emotional Debt & Being a Good AI Operator — with Anne Cantera
Summary
Anastassia and Anne Cantera discuss voice interfaces and key design principles for making voice chatbots "right." Anne designs voice experiences at scale, builds orchestration tooling for multi-agent systems, and teaches what she calls omnimodal design. Anne created VoiceofAI.io - a free AI upskilling resource for designers.
Key Takeaways:
Getting good AI output is a skill: provide requirements, design systems, and data so the model never has to make things up.
Never let AI write for you - let it interview you. That's the difference between AI-assisted and AI-generated — a thought innovator versus a slop generator.
The future is omnimodal design - technology that offers what you need, in the moment you need it.
Disclosure is non-negotiable. Voice AI can now listen and speak at the same time - making it easier than ever to fool people. Pretending a bot is human is a dark pattern.
Trust is gained in drops and lost in buckets. Companies that ship half-baked voice agents accumulate “emotional debt.”
Fast, easy, cheap: Voice projects are none of the three.
Replacing humans is backfiring. From flopped plans to outsource support to agents, to companies quietly rehiring.
Narrow domains win. A small, women-owned bank with one product can build a better chatbot than an enterprise with 110 products.
Bring in experts early — and learn to spot them. Leadership ego (“we don't need consultants”) makes messes that cost more to clean up later. But AI literacy comes first: an influencer with 500,000 followers and no AI work history is not an expert.
LLMs are a step, not the final frontier.
Chapters:
00:00 Intro: when the machines around us began to speak
02:24 Welcome Anne: dialing in from Long Island
03:26 From chef to conversation designer: an unusual path into AI
05:52 Self-taught in NLP: free chatbots, LexisNexis, and skills no school teaches
09:27 Omnimodal design: the perfect modality at the perfect moment
11:46 Being a good AI operator: spec-driven work and the interview technique
14:07 Human skills come before prompts
16:34 Voice tech today: machines that listen and speak at once
18:55 Ethics and dark patterns: why Anne left corporate
21:16 Emotional debt: trust is gained in drops and lost in buckets
24:42 Elastic organizations: core beliefs before AI design
26:59 Theory of mind: why machines cannot relate to us
31:39 Enterprise vs. everyone else: why small companies build better bots
36:42 Domains and jobs: machine translation, where nobody was laid off
39:42 Leadership ego: bring in your experts early
40:37 AI literacy first: wannabes, influencers, and real experts
43:49 The upskilling epidemic — and why VoiceofAI.io is free
49:20 Beyond transformers: LLMs are a step, not the final frontier
52:43 Robotics: industry first, our homes later
54:13 The AI class divide: haves, have-nots, and AI teachers
55:29 Closing: two more episodes promised — design and ethics
Hyperlinks:
Elementyl Intelligence Website
The Voice of Intelligence Design Handbook and Website
Anne Cantera Article on Conversational Systems
Anastassia Lauterbach - LinkedIn
First Public Reading, Romy, Roby and the Secrets of Sleep (1/3)
First Public Reading, Romy, Roby and the Secrets of Sleep (2/3)
First Public Reading, Romy, Roby and the Secrets of Sleep (3/3)
Weitere Episoden von „AI Snacks With Romy & Roby: Democratizing AI Technologies“



Verpasse keine Episode von “AI Snacks With Romy & Roby: Democratizing AI Technologies” und abonniere ihn in der kostenlosen GetPodcast App.








