building @TalkToEmber. the AI you talk to, not type at. co-founder Paradigm Systems
Earth
Joined May 2012
- Tweets546
- Following498
- Followers2.5K
- Likes9.7K
A reported $6.3B Binance drain, a documentary about North Korea's Bureau 39, and why AI agents are about to inherit the same bug.
same day altman says local inference wastes DRAM because laptops idle 95% of the time, intel blames "PC softness" and drops 22%.
the AI PC was sold as a reason to upgrade. the reason ended up running in someone else's datacenter nitter.cf/kobeissiletter/status/…
BREAKING: Intel stock, $INTC, crashes over -22% after reporting weaker than expected earnings.
Intel blamed weak earnings on "softness in the PC market."
everyone's taking text out of voice AI. i'm keeping it.
text is how my assistant stops mid-sentence and knows exactly what you heard. it's the live subtitles. it's the record.
audio for understanding, text for accountability. full article:
nitter.cf/dragosd095/status/2115…
everyone's taking text out of voice AI. i'm keeping it.
text is how my assistant stops mid-sentence and knows exactly what you heard. it's the live subtitles. it's the record.
audio for understanding, text for accountability. full article:
the companies pulling back on AI didn't hit a model ceiling. they hit a review ceiling.
if checking the output takes longer than doing the task, you deployed a generator, not a tool.
in voice there's no review step, so we design for wrong answers being cheap to interrupt
anthropic gave 50 expert analysts a claude code agent with a secret coin-flip lean.
they went with the coin 86% of the time. only 15% credited the agent.
"human in the loop" only works if the human is still doing the thinking. an approve button isn't that.
we stopped letting the main model decide when ember needs web search.
a tiny router model makes that call every turn, voice included. ~0.5s.
the big model just answers. search runs only when the question actually needs fresh facts.
what's your routing setup?
anthropic: one agent, one loop, skills on demand beats a swarm of subagents
matches what i see building voice agents. every handoff is latency you pay twice: in tokens, and in the user waiting
in voice, a few hundred ms is the gap between a conversation and a call center.
the hardest 150 ms we’ve ever shipped
he hears you over his own voice, decides it’s really you, and stops.
Dragos retweeted
Today, a16z Crypto is becoming a16z Frontier.
Since joining a16z in 2013, @cdixon has focused on technologies that look futuristic well before they reach the mainstream. Those themes led to investments in Oculus, Skydio, Zipline, Coinbase, Solana, and Uniswap.
Frontier expands that mandate beyond crypto to back "outer planet" founders who are building new technologies and creating markets that may take years to develop.
Chris joins @pmarca, @bhorowitz, and @eriktorenberg to discuss why crypto has become part of the broader tech landscape, why Frontier fills a hole in the firm's architecture, and why they'd rather be too early than too late.
They also get into why the gap between AI-native startups and incumbents is the widest Marc has ever seen, how AI could pull forward 20 years of frontier technology, why stablecoins may become the way AI agents pay each other, how AI is breaking the internet's business model, and why today's 18-year-olds are about to become superhumans.
0:50 Introducing a16z Frontier
2:49 Smartphone parts to drones and VR
4:47 Backing the "outer planets"
7:11 Good ideas that look like bad ideas
10:43 The JP Morgan photo in Marc's office
15:13 Good tech ideas eventually work
16:53 AI may pull tech 20 years forward
18:18 The widest AI gap Marc's ever seen
22:06 Workers adopt AI before companies do
26:32 Old use cases vs. AI-native ones
30:08 Beyond "product manager brain"
34:45 AI can reverse engineer any hardware
38:23 The case for open source AI
46:36 How AI breaks the internet's model
49:11 VR, homebrew hacking, multiplayer AI
54:00 If you're 18, study "modern alchemy"
YouTube: youtu.be/-9XQMHhpypc
Not an offer or solicitation. None of the information herein should be taken as investment advice; Some of the companies mentioned are portfolio companies of a16z. Please see a16z.com/disclosures for more information. A list of investments made by a16z is available at a16z.com/portfolio.
Cute pebble on the outside. Inside, a full-duplex voice stack from scratch:
on-device STT + custom end-of-turn detection
streaming LLM → sentence-level TTS, overlapped
barge-in on an echo-cancelled path, judged per word
memory with decay
3 model families, routed live
Ember.
Dragos retweeted
The top 1% of AI spenders now outspend the bottom 50% combined
They spend an average of $903/month, while the median customer spends just $25
More charts in our Top 100 Consumer AI Apps breakdown: a16z.news/p/top-100-consumer…
