1 of 11

NoteFish

Live calls across languages, in a voice you choose.

1

Hear the caller

2

Read captions

3

Hold to reply

2 of 11

Who we are

LEONARDO

Data Science & AI, Leiden

Daniel Rusu

Economics and Business Economics, Utrecht

  • Una Europa: real‑time AI disinformation detection, presented at KU Leuven
  • Research: Published Monte Carlo research at Moldova’s National Academy of Sciences at 17
  • LeLau: first encrypted offline phone AI, in iOS beta
  • 2nd place, Moldovan Science & Engineering Fair (MOLDSEF) 2024
  • EURHackNL finalist
  • Operator: Bootstrapped profitable multi-unit rental portfolio completely independently.
  • Dealmaker: Negotiated and structured complex high-yield lease contracts.
  • Product: Shipped zero-to-one digital platforms across multiple verticals.
  • Finance: Engineered rigorous venture-grade unit economics and models.
  • Execution: Scaled high-pressure operations from frontline facilities up.

3 of 11

Problem

2

people can share a call without sharing a language. NoteFish helps the agent understand and reply.

Your agent

speaks English

Your caller

speaks French

NoteFish

captions + voice

EN

Read captions in your language

FR

Hear replies in the caller’s language

First demo scenario: English-speaking agent and French-speaking caller. Live phone validation is the next milestone.

4 of 11

MVP

Caller

Open link

Tap Call

Speak French

Hear reply

French voice plays

Answer ›

Reply ›

Voice pipeline

OpenAI

Fish Audio

Browser

Agent

NoteFish

✓

Hear the caller and read translated captions

✓

Hold Space and speak in your own language

✓

Release to send speech in the selected voice

✓

End the call and keep the transcript

5 of 11

One agent, more languages

One agent

speaks English

Your voice

their language

French

First demo

Dutch

Selectable

German

Selectable

Spanish

Selectable

Italian

Selectable

Japanese

Selectable

Vietnamese

Selectable

Each language pair needs live testing

6 of 11

What powers NoteFish

Human Agent

Owns the conversation

OpenAI

Transcription and translation

Caller speech becomes captions

Listen

Reply

Agent speech becomes translated text

Fish Audio

Speech in an approved voice

Select

​

Enrolled or licensed

Speak

�French reply

Hold to speak, release to send

Control

�You choose timing

Review

�Keep the transcript

7 of 11

First customers

Proposed pilot segments

01

Service Desks

Local service teams taking calls from residents in several languages.

Use case: repair intake

02

Hospitality

Hotel teams helping international guests with requests in a language the agent does not speak.

Use case: guest support

03

Customer Support

Small support teams serving customers across borders without a specialist for every language.

Use case: order help

8 of 11

How we are different

A human-led call, with translated words and an approved voice

01 / CAPTIONS

Hear and Read

Hear the caller’s original audio while reading captions in your language.

02 / CONTROL

Hold to Speak

Hold Space to record your reply. Release to translate and send it.

03 / VOICE

A Chosen Voice

Fish speaks the reply using your enrolled voice or a licensed voice.

04 / CONSENT

Approved Voices

Enroll voices off-call. Never clone a caller or someone on the live line.

05 / RECORD

Call Transcript

Keep source text and translations together when the call ends.

9 of 11

Proposed Pilot Model

CUSTOMER

Small service teams

Teams Handling calls across languages

PRICING

Hypothesis to test

Seat + use Subscription plus audio usage

PILOT

Target: 3 teams

3 teams English + French

SUCCESS

Measure on real calls

Resolve Useful replies

10 of 11

Where this can go

The next milestone is a complete conversation on a real phone.

schedule

DEMO

Browser calling

A phone opens a private link. The laptop answers with captions and Fish replies.

call_split

NEXT

Real team pilots

Test call completion, reply accuracy and latency with English and French before expanding.

language

LATER

More languages

Validate more language pairs and add optional phone-number calling after the browser workflow proves useful.

11 of 11

Our Ask

3

pilot teams

Initial language pair

English + French

Proposed Pilot Plan

groups

Service Team Partners

Recruit teams that handle recurring calls across languages.

dns

Live Call Validation

Test phone audio, translated captions and Fish replies end to end.

gavel

Voice Consent

Use enrolled or licensed voices with clear permission.

monetization_on

Pricing Feedback

Measure usage and willingness to pay during the pilot.