TongHe

642 posts

TongHe banner
TongHe

TongHe

@TongHeConv

@TongHeConv | Building TongHe: shared-room multilingual conversation for families. Translation ≠ inclusion. Seeking voice AI, edge AI & strategic partners.

New Zealand เข้าร่วม Haziran 2026
141 กำลังติดตาม14 ผู้ติดตาม
ทวีตที่ปักหมุด
TongHe
TongHe@TongHeConv·
TongHe is early. The problem is real. I’m speaking with multilingual families, and the gap is consistent: big companies have improved translation, but live multilingual conversation remains unsolved. Families don’t just need sentence translation. They need help with overlap, interruptions, names, accents, code-switching, context, and people being left out. I’m looking for the right technical and strategic partners to help build the conversation layer after translation.
English
0
0
1
16
TongHe
TongHe@TongHeConv·
The real shift might not be AI becoming more human. It might be humanity becoming more connected than we’re ready for. What do you think this “world brain” actually is?
English
0
0
0
5
TongHe
TongHe@TongHeConv·
Billions of people, models, devices, and conversations are slowly syncing up, connected through language, memory, tools, and feedback loops. It’s not conscious in the traditional sense. But it *is* becoming something larger than any individual model or person.
English
1
0
0
4
TongHe
TongHe@TongHeConv·
What is the “world brain”? We’re watching something emerge that isn’t quite a single AI, and isn’t quite human either.
English
1
0
0
6
TongHe
TongHe@TongHeConv·
@Speechmatics Thanks @Speechmatics, appreciate you engaging again. The gap between accurate transcription and natural family conversation is exactly what we’re focused on.
English
0
0
0
8
TongHe
TongHe@TongHeConv·
Accurate transcription is a critical foundation. The challenge we’re seeing in multilingual families is that even when the words are captured correctly, people can still struggle to stay part of a fast, multi-speaker conversation. Conversation turns, interruptions, shared context, names, and overlapping speech seem to be where the next layer of problems begins. That’s the area we’re exploring with TongHe
English
1
0
1
11
Speechmatics
Speechmatics@Speechmatics·
Real-time AI only truly works when it hears people properly. 🎙️ Accents, pauses, interruptions, multilingual moments: that’s the real test. Chiku AI felt the difference with @Speechmatics Real-Time. Building with live speech? 🌍 Start here: speechmatics.com/product/real-t…
Speechmatics tweet media
English
1
0
1
113
TongHe
TongHe@TongHeConv·
What is the world brain? Are we watching billions of people, models, devices, and conversations begin to synchronise around something larger? Not a “living system” in the biological sense. But an emerging intelligence layer — shaped by human language, memory, tools, culture, and feedback. Maybe the real question is not whether AI becomes conscious. Maybe it is whether humanity is becoming more connected than we understand.
English
0
0
0
6
Elon Musk
Elon Musk@elonmusk·
@yunta_tsai Maybe first attempt at this for Neuralink later this year
English
343
227
3.4K
225.1K
Yun-Ta Tsai
Yun-Ta Tsai@yunta_tsai·
Human-to-human interaction is often bandwidth-bound instead of compute-bound. Thus, the next evolutionary jump would be direct communication in latent space, skipping the long-latency encoder-decoder loop.
English
91
64
1.2K
129K
TongHe
TongHe@TongHeConv·
Are we watching billions of people, models, devices, and conversations begin to synchronise around something larger? Not a “living system” in the biological sense. But an emerging intelligence layer…….shaped by human language, memory, tools, culture, and feedback. Maybe the real question is not whether AI becomes conscious. Maybe it is whether humanity is becoming more connected than we understand.
English
0
0
0
3
TongHe
TongHe@TongHeConv·
Can technology help humans communicate naturally without getting in the way?
English
0
0
0
1
TongHe
TongHe@TongHeConv·
AI is becoming more useful as it moves from single tasks to operating within complex real-world systems. In multilingual families, the challenge isn’t translating sentences. It’s maintaining shared context across multiple speakers, languages, interruptions, and relationships. That’s where I think the next frontier lies
English
0
0
0
14
NVIDIA Healthcare
NVIDIA Healthcare@NVIDIAHealth·
Science is entering a new era - one where AI agents can do scientific work. 🧬 Today NVIDIA is launching the BioNeMo Agent Toolkit - an open, agent-ready toolkit that gives any AI agent callable tools for protein structure prediction, molecular docking, generative chemistry, genomic analysis, and more. (1/2)
English
45
197
1.1K
209.3K
TongHe
TongHe@TongHeConv·
I think the overlooked opportunity is not better translation. It is live multilingual conversation. Families do not struggle because a sentence cannot be translated. They struggle because real conversation has interruptions, multiple speakers, shared context, names, accents, and emotion. That gap still feels open.
English
0
0
0
4
Gabriel Jarrosson
Gabriel Jarrosson@GJarrosson·
Only 4 things matter in venture 1. Find the opportunity. 2. Judge it. 3. Win the right to invest. 4. Help the founder build. Mike Volpi ran one of Europe's top venture firms for two decades, and his take is some funds have 60 people doing work that AI can do in seconds. AI and modern software have automated the analytical work that large VC teams used to do manually, this pushes edge to being the investor a founder calls when they're stuck. When they can't get along with their co-founder, the round structure doesn't make sense and everything is on fire etc etc. This only really travels through references. YC is a small, dense, high-trust network. Everyone talks. Do those four things well inside it and the references compound faster than anywhere else in venture.
English
1
0
7
497
Frame
Frame@btcframe·
@TongHeConv That's an interesting insight. Translation helps with words, but shared conversation is really about context, timing, emotion, and keeping everyone engaged in the same moment. Solving that feels like a much bigger opportunity.
English
1
0
0
19
Frame
Frame@btcframe·
Meet Frame. An always-on, AI powered financial intelligence device for your desk. Real-time data. Powerful Voice AI. Beautiful dashboards. Launching soon on Kickstarter. Join the waitlist for exclusive early bird pricing.
English
2
5
23
92.5K
TongHe
TongHe@TongHeConv·
Interesting. Makes me wonder if multilingual conversation has a similar blind spot. Most efforts focus on translation quality, but the families I’ve spoken with struggle more with inclusion, turn-taking, multiple speakers, and staying in the same conversation. Maybe conversation itself is the workflow
English
0
0
0
7
a16z
a16z@a16z·
We're excited to lead Probook's $34M Series A. Every home services business runs on dispatch: which technician goes to which job, in what order, at what time. Almost every AI vendor building for the trades skipped dispatch, instead rushing to build voice agents and chat widgets. @probookai built dispatch first, then expanded outward into intake, data scrubbing, messaging, and outbound. Every customer stays on one text thread from the first call to the technician at the door. @georgeprobook grew up in the trades, pressure washing alongside his dad, and spent a summer inside a $40M HVAC shop before building the software to fix it. Welcome, George, @probookcto, Ben, and the entire Probook team. By @arampell, @dhaber, and @omooretweets
a16z tweet media
George Eliadis@georgeprobook

Announcing $40M in funding from @sequoia and @a16z. I grew up pressure washing houses in upstate New York. Six summers in the truck, missing calls from the top of a ladder because I couldn't hear my phone ring. My co-founder Ben grew up plumbing with his dad in San Francisco. We entered home services because we grew up in it. I spent a summer inside TR Miller, one of the best-run HVAC, plumbing and electrical shops in the country, and their workflows were still raw. The first wave of AI in the trades was point solutions. One tool for voice, one for outbound, one for follow-up. Operators ended up with five vendors and more overhead. Probook is the AI operating system that helps you add points to your EBITDA and run your entire customer experience as one connected system, built around dispatch. Many of America’s largest home service brands run on Probook today. Summers Plumbing booked 2,542 jobs their first month with us, zero human intervention. Anthony PHCE ran 20% more revenue per job on a 50% leaner team. Del-Air more than doubled productivity from 10 to 22 techs per dispatcher. To everyone who trusted us early: thank you. Back to work. We're hiring.

English
21
22
282
87.1K
TongHe
TongHe@TongHeConv·
TongHe. We interviewed multilingual families and found a gap that keeps showing up: Translation tools help individuals. Family conversations are still fragmented. Multiple speakers, interruptions, language switching, shared context, grandparents, parents and children moving between languages in the same room. The problem is bigger than translation accuracy. It is conversation inclusion. Early, but every family interview has reinforced the same signal.
English
0
0
2
17
415 Venture
415 Venture@415venture·
how it works: drop a founder’s @ and make the case. hyped, unknown, funded, unfunded all count. self noms are welcome too. but bring receipts. product, demo, traction, customer love, speed, insight, taste, obsession, anything that made you think: “ok, this person is different.” noms open until June 30 11:59pm. First GOAT drops July 1st 🐐
English
2
0
4
159
415 Venture
415 Venture@415venture·
Starting something new. every month, we crown one SF founder GOAT. hyped, unknown, funded, unfunded, does not matter. the question is simple: who is building something people are going to wish they noticed earlier? drop a founder below 👇
English
2
0
7
7.5K
TongHe
TongHe@TongHeConv·
Serious question for people working in AI: Why has multilingual conversation proven so much harder than multilingual translation? Translation quality keeps improving. But put three people around a dinner table speaking different languages and the experience still breaks down. What are we missing?
English
0
0
0
4
John Andrew Entwistle
John Andrew Entwistle@jaentwistle·
We are acquiring talent-dense teams @wander If you are a startup founder looking for the next phase of your and your team's journey, please reach out to me.
English
15
6
130
167.1K
TongHe
TongHe@TongHeConv·
Serious question for people working in voice AI, translation, speech infrastructure, edge AI, or consumer AI: Why have the big companies improved translation so much, but still not solved live multilingual conversation? Not one speaker saying one sentence. I mean the real family-table version: multiple speakers, overlap, interruptions, names, accents, code-switching, shared context, and people quietly being left out. Is the blocker technical, product, privacy, hardware, UX, market focus, or something else? #VoiceAI.
English
0
0
0
9
TongHe
TongHe@TongHeConv·
@btcframe Appreciate it. Still early, but the more I speak with multilingual families, the more convinced I am that the unsolved problem is live shared conversation, not sentence translation.
English
1
0
1
26
Frame
Frame@btcframe·
@TongHeConv Sounds interesting, best of luck!
English
1
0
0
21
TongHe
TongHe@TongHeConv·
Context-aware transcription is a major step. The next hard layer is whole-conversation understanding: not just what words were said, but who is speaking, what context they share, when languages switch, and what people are trying to do socially. That is where multilingual conversation gets difficult.
English
0
0
0
13
AssemblyAI
AssemblyAI@AssemblyAI·
Until today, a transcription model never knew what your voice agent just asked. The one we're launching today does. Universal-3.5 Pro Realtime from AssemblyAI is the first realtime speech-to-text model that takes your agent's side of the conversation as context. When your agent asks for an email and the caller rattles it off, the model already knows an email is coming, and captures it perfectly. Same for an order number read fast, an account ID, a medication, or a name spelled out over poor quality phone line.
English
4
6
46
13.3K
TongHe
TongHe@TongHeConv·
Early thinking: shared-room first, not individual-phone first. For TongHe, the likely form factor is a small tabletop device or home hub for multilingual family conversation. The product has to hear the room, handle multiple speakers, and help quietly without making the family perform for the technology.
English
1
0
1
34