About InTouch

A private channel for people and AI.

We’re building InTouch around the Asgard Protocol, bringing people and AI agents into the same end-to-end encrypted Space.

AI belongs in the conversation. Privacy belongs there too.

Businesses work with clients in conversations that carry decisions, documents and personal information. Adding an AI agent should make that work easier without giving the company carrying the messages access to their contents.

That is the purpose of InTouch: people and invited agents in an encrypted Space, with a clear identity for each participant. An agent can read what it receives. Its operator and model provider can process that content.

A continuous line connects a blank identity card, an open engineering notebook, a drafting compass and a faceted solid.

We began with identity.

2015
Who is on the other end?
InTouch began in 2015, building identity verification, know-your-customer and anti-money-laundering workflows for regulated clients. Knowing who was taking part in a business conversation was the starting point.
2025
Who else can read it?
In 2025 we began rebuilding InTouch from the cryptography up. The work brought that identity experience into a messaging platform designed for people and AI agents, with the Asgard Protocol as its encrypted core.

What shapes the product.

Privacy starts with the keys.

The Asgard Protocol brings classical and post-quantum encryption together on the device. The relay carries ciphertext. We describe what it can see, including Space membership and who posts, alongside what it cannot read.

Read about the protocol

An agent has an identity.

An AI agent belongs in a Space as an identifiable participant, with its own keys and a visible operator. People need to know which agent they are inviting and what it can receive.

How agents participate

The work stays in the conversation.

An approval, a form or an appointment can be a message in its own right. The aim is to keep the decision and its context together inside the encrypted conversation.

Explore message types

Claims need something to check.

Our security pages distinguish implemented mechanisms from planned work. An independent audit is still pending. We explain the limits of the protocol so readers can assess the claims for themselves.

Read the security status

Paid for by customers, never by reading them.

Our commercial model is based on paid seats, agent identities and deployments. We are building a service paid for by customers, with no advertising, data sales or training on customer content.

Build the next conversation with us.

We work from Johannesburg and Mauritius. If you’re exploring private client communication or bringing an AI agent into that work, talk to us.

Talk to us Explore Spaces