Everything can be faked.Except one thingHuman Presence

Not who, but if

The problem
Anyone can fake a voice
in seconds, and the copy is good
Why it matters
A safer tomorrow
evidence in front of the action, not after
Deeptech from Amsterdam
SCROLL
First live observation
Awaiting
Date
Context
Hypothesis
Human and Agent Presence can be shown without identity or contentstated 19 November 2026
Observation
Evidence
Novelty
Limitations
Engine
Validated
Real-time Continuity™ v0.20, cross-platformvalidated release 20 August 2026
Measured live
Public air traffic stream
Integrations
Live
Call center
In preparation
Enterprise voice
Research
Agent-to-agent
The technology

Real-time Continuity™

Is this still the same interaction

measured in the near past, while it runs

The first call counts as much as the thousandth

Real-time Continuity Infrastructure™·the temporal evidence layer
Human Presence Infrastructure®·the entrypoint·not who, but if
The problem

A voice is no longer proof

Anyone’s voice can be copied in seconds, and the copy is good. The one thing people still trust on the phone is now the easiest thing to fake.

At the same time, AI agents are starting to act. They book, they transfer, they approve. They check who is there at the start of a call, and then take that on faith for the rest of it.

Intono develops a new technology: Real-time Continuity. It measures coherence in a call, without knowing who you are and without reading what you say: is this still the same agent, and is a human still present. Not once at the start. Continuous. It flags the moment that coherence breaks.

Left: one check at the start of a call, and then the call is hijacked. Right: verified human presence at 0, 30, 60 and 300 seconds, with no evidence and no action at 120 seconds.
Illustration of the mechanism · not a live measurement
On record

25 million dollars, approved in a video call where every other face was generated

A global engineering firm · Hong Kong, 2024 · confirmed by the company

129 million dollars, wired by employees who were doing their job

Six companies · United States, 2026 · a cloned executive voice, then a second call from a fake lawyer · federal indictment, charges and not convictions

The arms race

Intono looks past the perfect voiceNow, and tomorrow

How good the copy sounds is not a question Intono asks. The measurement does not get weaker when the next generation of voices gets better.

Time is the only witness

Where this goes

Voice becomes ordinary, trust does not

Three steps, and the third one changes what has to be proven.

Now
AI speaks with people

Service desks, reception, sales, support. People still know there is a machine on the line.

Two to five years
AI acts for people

Call my insurer. Negotiate the energy contract. Ask for a quote. The agent holds the conversation on its own, and the call becomes a transaction.

Five to ten years
AI speaks with AI

Your agent calls the airline, an agent answers, and the two settle it between them. You get the outcome. The voice becomes a protocol.

When every voice can be perfect, recognising the voice proves nothing. One question is left, and no voice can answer it.

Why was this conversation allowed to lead to this action?

That is the question Intono was built to answer

Who pays, and for what
The party that acts on the call
A consequential action needs its own fresh proof, not the trust the call started with
The platform that supplies the agent
Its agent can be shown to be bound, and the human present, per call
The party that has to answer for it
What held during the call can be stated afterwards, without a recording

When the proof is not fresh, nothing is blocked·the call goes to a human

The first step

Start beside what you already run

The first engagement is a shadow pilot. Intono watches one bounded class of action and verifies it live, keeping no recording. It recommends, it does not execute: no downstream effect in your systems, nothing to roll back.

Scope
One bounded class of action. Shadow and recommend only, no execution rights.
Integration
A hosted endpoint, called from your side over HTTPS with a key issued for the trial. No inbound access to your systems.
Package
API contract, worked examples, runbook, checksums. Handed over before the first call.
Data
No recording kept, no content read, no identity resolved.
Term
Eight weeks, fixed scope and fixed term, agreed in writing before the start.
Integration package complete
Shadow mode · no execution rights

We left the arms race

Status

Early, deliberately quiet, now coming forward

Intono has been built quietly for over a year. The architecture is patent pending. The integration package for the first enterprise pilot is complete.

Filed patent portfolio
Eight patent families filed, two trademarks
8
Patent families
Filed 2025 · 2026 · patent pending
2
Trademarks
Human Presence Infrastructure® · Real-time Continuity Infrastructure™
Technical detail stays with counsel until the applications publish
Group
Intono AI B.V. is part of a privately held group in Amsterdam.
Ownership
The patent families are held by mAvI Innovation B.V., the group’s IP company, and licensed to Intono. Ownership sits apart from operations.
Operating history
Gallery Play®, in the same group, runs a B2B music streaming platform for commercial spaces across Europe, on proprietary technology. In operation since 2015.
Operations
Day to day operations, support and administration run on the group’s team in Amsterdam, shared across its companies.
First enterprise pilot in preparation
Amsterdam

The last thing you can’t fake

Get in touch

Work is shared under NDA

contact@intono.ai

Michiel Borstlap · Inventor Real-time Continuity & Founder

Technical material follows under NDA; the integration package can be handed over once a named technical contact and a secure channel are agreed.