Not a chat bot. A companion.

A character, not a setting.

Built from behavioural research on how people talk, so it feels natural rather than scripted.

In its own voice, in whichever language the person opens with, South African ones included.

Tell it once.

It keeps a memory for each person: interests, context, how they like to be spoken to.

In time, everyone is talking to a slightly different companion.

It speaks first. It also knows when not to.

It texts first when there is a reason, and goes quiet when you would. The one it had at two in the morning is never sent.

Keen, not needy.

Multi-agent layer.

Underneath the companion sits a layer of specialist agents. When an answer needs real knowledge, one of them goes and fetches it.

The companion answers from what comes back, not from what it might have guessed.

It sends anything.

Not only words. A photo, a video, a document, whatever the answer actually needs.

It goes both ways: send it the thing that is broken, and it will look.

Voice.

Send it a voice note. It listens, and it can answer in a voice of its own.

Being read to is not the same as being spoken to.

Every conversation leaves a layer.

What people ask for, what they mean, what they would never tell a form.

It arrives as data your team can query, not transcripts to read.

Talk is not the whole job.

Mid-sentence, it can take a person to the shop, to their rewards, or to your team as a lead.

The points are set by what the person needs, not by a menu.

One system. Your APIs, your character.

One codebase, kept and improved by us.

You plug in your APIs and a character. The rest is already built.

Fit the skills the job needs.

Reading a label, recognising a product, whatever the character should be able to do.

Each one seats on the same apparatus. The rest of the system does not notice.

Scope sets the cost.

Set-up and token cost both follow the scope: the skills fitted, the systems touched, the volume carried.

We put numbers to both once the full scope is agreed.

Agencies can resell the platform to their own clients. Their revenue sits in the same two places: the set-up, and the recurring monthly token cost.