Working Prototype

Prototype demo

VOZURA

Conversation Intelligence that remembers what matters.

Don't just listen.Understand.

AI can transcribe a conversation in seconds. The real value starts when it understands the conversation in context — across every conversation that came before it.

  1. Voice
  2. Memory
  3. Intelligence
  4. Action

Founder & AI Product Manager · Working Prototype · AI-assisted development

Don't just transcribe conversations. Understand them.

Working Prototype — In active development

VOZURA is not a launched customer product. The demo uses synthetic, fictional data and contains no real customer metrics.

Read the full Product Management case study

How VOZURA thinks

From one conversation to a decision.

Scroll through the concept: a customer call becomes structure, memory and a recommended next step.

  1. 01

    Conversation

    A real customer call — tone, pauses and pressure that a CRM note never captures.

    00:12:41CustomerAccount lead
  2. 02

    Understanding

    Speech becomes text with speaker separation, and sentiment, intent and objections are read in context.

    Transcript

    Customer…the rollout keeps slipping on our side.
    Account leadLet's fix the sequence before the next milestone.
    CustomerWe need clarity on the timeline first.
    SentimentObjectionCommitment
  3. 03

    Memory

    The current interaction connects to previous conversations, so recurring topics stop starting from zero.

    PreviousMemoryToday
    Q1 callSupport threadRecurring topic
  4. 04

    Decision intelligence

    Patterns surface where a relationship is at risk — and where value can grow.

    Risk

    Opportunity

  5. 05

    Next best action

    The output is not a summary. It is the next step, with the reason behind it.

    Next action

    Confirm the timeline in writing before the next milestone.

    Reasoning

    Recurring topic · Objection · Risk

Open the VOZURA demo

Concept and prototype. Synthetic demo data.

Working Prototype

Demo data — fictional customer

Demo scenario

Acme AG

Jenia JostHead of Operations

Step 1 — Context

The conversation didn't start today.

Before analysing today's call, the concept loads what this customer said in previous conversations.

  1. 3 months ago

    Our SAP integration is still causing problems.

    AI memory · Recurring issue detected

  2. 6 weeks ago

    Can you make sure this gets resolved before our next review?

    AI memory · Promise detected

  3. 2 weeks ago

    We haven't seen much progress.

    AI memory · Relationship risk increasing

  4. Today

    New conversation available

Today's conversation

Quarterly operations review

Recorded18:42
  • Jenia — Customer

    We've talked about the SAP integration several times now. The problem is still affecting our operations, and honestly, I'm getting frustrated because we were told this would be resolved by now.

  • Sales Manager

    I understand. Let me check where we are with the technical team and make sure we get a clear update.

  • Jenia — Customer

    I really need a concrete answer this time. We can't keep discussing the same problem every few weeks.

Demo data — fictional customer

Conversation Intelligence

Conversation Intelligence

VOZURA extracts structured behavioral and commercial signals from conversations instead of stopping at transcription.

Speaker diarisation

Separates who is speaking, turn by turn.

Talk / listen ratio

Measures the balance between talking and listening within a call.

Monologue detection

Flags long, uninterrupted stretches of talk time.

Interruption rate

Tracks how often speakers interrupt each other.

Sentiment timeline

Follows how customer sentiment shifts across a conversation.

Intent / interest detection

Reads signals of customer interest or intent within the dialogue.

Focus detection

Identifies which topics a conversation actually concentrated on.

Objection detection

Flags moments where the customer raises a concern or objection.

Competitor keyword detection

Detects mentions of competitor names or products.

Question rate

Tracks how many questions each speaker asks.

Customer Memory

Customer Memory

VOZURA connects signals across conversations so customer context does not reset after every call.

Recurring issues

Surfaces problems a customer has raised more than once.

Promises / commitments

Tracks commitments made in a conversation and whether they were kept.

Blockers

Highlights unresolved obstacles standing in the way of progress.

Momentum

Reflects whether a customer relationship is trending up or down over time.

Decision-maker linkage

Connects conversation signals to the relevant decision-makers.

Persistent customer context

Carries context from one conversation into the next, instead of starting from zero.

Differentiation

From transcription to intelligence.

VOZURA is not only a transcript or a summary. It is built around persistent context that carries across conversations, not just within one.

Structured signals — relationship momentum, objections, intent — are tracked over time, so a pattern across five calls is visible, not just what was said in one.

Traditional conversation intelligence

  • Transcript
  • Summary
  • Keywords

AI Voice Intelligence

  • Conversation
  • Context
  • Memory
  • Patterns
  • Business impact
  • Recommended action

My Role

My Role — AI Product Management

I shaped VOZURA from the initial customer problem into a working product concept, making the product decisions behind what to build, what to prioritise and how the experience should work.

AI Product Manager with hands-on, practical technical product work — not a software engineer.

Read the Product Management Case Study

Next step: validate with real users and customer discovery.

Users

Sales · Customer Success · Account Management · Support

Segments

B2B SaaS · Telco · Industry / Enterprise

AI / Tech

AI / Tech

AI / Product Layer

  • LLM workflows
  • Transcription
  • Conversation analysis
  • Persistent memory

Tools / Infrastructure

  • Deepgram
  • Supabase
  • Claude Code
  • GitHub

Development Approach

  • Structured QA
  • E2E testing
  • AI-assisted development

Language

Swiss German transcription work / beta exploration

How it flows

  1. Voice
  2. Transcription
  3. Understanding
  4. Memory
  5. Reasoning
  6. Action

Product vision at scale

One conversation is a signal.The vision: thousands become intelligence.

Conceptual demo data

127

Conversations analyzed

18

Recurring issues detected

7

Unresolved product problems

4

Emerging customer risks

Illustrative example, not real customer data or usage.

Emerging product problem

SAP integration

Detected across multiple fictional customer conversations.

Signal

Increasing

Potential action

Investigate product / technical root cause.

Current Status

Working Prototype — In active development

VOZURA is not a launched customer product. It has no paying customers, no revenue and no employees. The demo uses synthetic, fictional data. Real-user discovery and customer validation are the next step.

Kayky AI

Ask Kayky AI about this product

The assistant answers from the same verified knowledge base used across this site.

What if every customer conversation could become product intelligence?