TongHe Human Connection

Founder @TongHeConv | Tabletop AI device for natural multilingual family conversations. One table.Every voice.

[TongHe Human Connection @TongHeConv]

TongHe is a human-first multilingual conversation system that helps families and small groups stay naturally in the same conversation across languages.

Problem

Translation tools are improving fast, but real conversation is still hard.

Around a family table, people interrupt, overlap, use names, accents, jokes, shared context, code-switching, emotion, and silence. Existing tools translate words, but they do not yet support the flow of human conversation.

First use case

The first wedge is the multilingual family table: Mandarin-English, multi-speaker, always-on, low-friction, privacy-conscious conversation support.

Why now

Voice AI, multilingual speech recognition, edge AI, and consumer hardware are finally close enough to make this possible. The missing piece is not only translation quality. It is conversation design.

Ask

I am looking for a technical cofounder or early strategic backer who understands voice AI, multilingual speech, consumer AI hardware, or real-time conversation systems.

The goal is to turn TongHe from a clear human need into a focused prototype and fundable early company.

[TongHe Human Connection @TongHeConv]

@zjearbear @ArgaLabs @superset_sh @totalistrading @heyclicky @archer_money @manicule @sazabi @memorydotstore One table. Every voice.

[TongHe Human Connection @TongHeConv]

I’ve now spoken with 10 Mandarin-English families about dinner-table conversation.

The pattern is clear:

The problem is not sentence translation.

It is being excluded from fast, overlapping family conversation.

Existing tools are too individual, too interruptive, or too awkward for the table.

TongHe is being designed around shared-room conversation, not just translation.

[TongHe Human Connection @TongHeConv]

Multilingual family conversation around the dinner table.

My wife is Chinese, and many family conversations happen naturally in Mandarin.

The challenge isn’t simply translation.

It’s being able to participate fully when multiple people are talking, stories are being shared, opinions are exchanged, people interrupt each other, and conversations move quickly between speakers.

In many Chinese families, communication around the dinner table matters. Family history, humour, advice, debate, and relationships are all built there.

I don’t just want a translation of the conversation.

I want to be part of the conversation.

Translation is getting remarkably good.

The harder frontier is helping people communicate naturally across languages without losing the flow, timing, context, and human connection that make conversations meaningful.

[TongHe Human Connection @TongHeConv]

Multilingual family conversation around the dinner table.

My wife is Chinese, and many family conversations happen naturally in Mandarin.

The challenge isn’t simply translation.

It’s being able to participate fully when multiple people are talking, stories are being shared, opinions are exchanged, people interrupt each other, and conversations move quickly between speakers.

In many Chinese families, communication around the dinner table matters. Family history, humour, advice, debate, and relationships are all built there.

I don’t just want a translation of the conversation.

I want to be part of the conversation.

Translation is getting remarkably good.

The harder frontier is helping people communicate naturally across languages without losing the flow, timing, context, and human connection that make conversations meaningful.

[TongHe Human Connection @TongHeConv]

@googledevs Low latency is critical here.

In multilingual conversation, delay is not just a technical metric; it changes the social rhythm of the room.

The real test is whether people can keep participating naturally while language is being bridged in real time.

[TongHe Human Connection @TongHeConv]

This is a big step.

One-to-many translation is powerful, especially for presentations and global audiences.

The next hard problem is many-to-many conversation, multiple speakers, interruptions, accents, names, code-switching, timing, and shared context.

That is where multilingual AI gets really interesting.

[TongHe Human Connection @TongHeConv]

@DanielSmidstrup Building TongHe, a tabletop AI conversation device for multilingual families. Translation is getting solved quickly. The harder problem is handling natural conversation: interruptions, code-switching, names, accents and shared context

[TongHe Human Connection @TongHeConv]

This is a really important area.

Idioms are where literal translation often breaks down because the meaning depends on culture, context, and how people actually speak.

For multilingual AI, the challenge is not only understanding words across languages.

It is understanding communication across communities.

[TongHe Human Connection @TongHeConv]

This is a strong direction.

Speaker separation, language identification, accented speech, names, and context are not minor details, they are where real-world multilingual conversation succeeds or fails.

The next frontier is not just speech-to-text accuracy.

It is keeping people naturally in the conversation.

[TongHe Human Connection @TongHeConv]

Congrats: strong momentum.

Voice-first AI feels like the right direction because conversation is the most natural human interface.

The hard part now is making it work in real-world messiness: timing, interruptions, accents, names, context, and people speaking the way they actually speak.

[TongHe Human Connection @TongHeConv]

The expressiveness controls are interesting.

Voice AI is moving from “can it answer?” toward “can it participate naturally?”

Pace, tone, timing, interruptions, accents, names, and context are going to matter as much as raw intelligence.

Conversation is a human interface, not just a model output.

[TongHe Human Connection @TongHeConv]

The multilingual interview simulator is a strong example.

What interests me is where this goes next: from structured interview flows into messy real conversation.

Real multilingual communication has interruptions, accents, names, code-switching, timing, and shared context.

That is the harder frontier.

[TongHe Human Connection @TongHeConv]

This is an important direction.

Language switching mid-conversation is closer to how real multilingual people actually communicate.

The harder step is making that work naturally in messy group settings too, interruptions, overlapping speech, names, accents, timing, and shared context.

[TongHe Human Connection @TongHeConv]

@aiseomastery This is a big signal.

Multilingual voice is moving fast, but the next challenge is not just adding more languages.

It is making conversation feel natural across accents, interruptions, names, code-switching, and shared context.

That is where the real product frontier is.

[TongHe Human Connection @TongHeConv]

That is why I keep coming back to this:

Human first.

Communication first.

Translation second.

The goal is not just better translation.

The goal is helping people stay in the room.

[TongHe Human Connection @TongHeConv]

Multilingual AI should not just ask:

“What did they say?”

It should ask:

“How do we keep everyone in the conversation?”

Translation matters.

But participation matters more.

[TongHe Human Connection @TongHeConv]

One overlooked part of multilingual communication is cognitive load.

The person who understands least often works hardest.

They are translating, guessing context, watching tone, managing timing, and deciding whether to interrupt.

That is a participation problem.

[TongHe Human Connection @TongHeConv]

As a founder working on multilingual conversation, I keep coming back to one thing:

Translation is not the whole problem.

The harder problem is helping people stay in a natural conversation when languages, accents, timing, names, interruptions, and family context all collide.

That is where I think multilingual AI has to go next.

Not just “what did they say?”

But “how do we keep everyone in the room?”