Wednesday, October 7, 2026ArchiveSearchAsk the paper

The Computomatix Times

All the posts fit to save — curated from @computomatix's bookmarks & likes on X

Edition of Sunday, March 29, 2026

3 stories

Product Manager Says Builder Role Replaces Traditional PRD Work

Moe Ali relays an account of an AI product manager at a frontier lab who no longer writes PRDs, instead prototyping with Claude Code, running evals and shipping reference implementations. The post argues PMs should act as builders with agents.

Original post · 2 min read
Just talked to an AI Product manager making $375K at a frontier lab.

She hasn't written a PRD in 8 months.

Not "she uses AI to help write them faster." She has not opened a PRD template in eight months because she just doesn't need it anymore.

Her day looks nothing like what PMs do:

She wakes up, opens Claude Code, and has a working prototype running before her first meeting of the day (not a wireframe or a figma mockup someone needs to hand off to an engineer). A testable version of the idea - built/shipped by her in the same morning.

While that prototype is running, she's pulling model outputs and running evals. She knows what hallucination looks like in her specific use case. She knows what latency threshold breaks the user experience. She knows the token cost per query and what that means for margin at scale.

She reasons about infrastructure the way a CFO reasons about a P&L. When something needs to be built for real, she doesn't go write a ticket and wait two sprints. She ships the first version herself. hands it to engineering as a working reference implementation, not a requirements doc full of edge cases nobody reads.

The meetings she's in aren't about alignment. They're about what's already shipped and what's blocking the next thing.

Her mental model isn't:

- "manage the roadmap."
- "be the voice of the customer."
- "facilitate cross-functional collaboration."

Those aren't wrong exactly, they're just from a different era.

The mental model that gets you to $480K at a frontier lab in 2026 is simpler and harder at the same time:

- You are the builder.
- The agents are your team.
- Your job is to ship.

She said the output gap between PMs who operate this way and PMs who don't is already 3 to 4x. And this is inside a lab where literally everyone around her is working the same way.

Here's the thing nobody wants to say out loud: the "I write specs and run standups" PM isn't being replaced by AI. The job isn't disappearing into a chatbot. It's being absorbed by the PM sitting two desks over who stopped waiting for engineers and started building herself.

Really crazy how fast the job description changed and really crazy how few people have noticed.
♥ 1.1K · ⟲ 75 · 👁 193.9KView on X ↗
AI7/10

Microsoft Re-Releases VibeVoice Open-Source Voice AI With Safeguards

Microsoft Re-Releases VibeVoice Open-Source Voice AI With Safeguards

Nav Toor describes Microsoft's open-source VibeVoice, claimed to clone voices from 10 seconds of audio, generate 90-minute multi-speaker audio, and transcribe with speaker labels. He says Microsoft pulled the repo over deepfake misuse and re-released it with watermarks and safety controls, under an MIT license.

Original post · 2 min read
🚨 Microsoft just open sourced a voice AI that was too dangerous to keep live.

They took it down. Added watermarks and safety controls. Then re-released it. For free.

It's called VibeVoice.

Microsoft's frontier open source voice AI.

Clone any voice from 10 seconds of audio. Generate 90 minutes of multi-speaker conversation. Real-time streaming. All running locally on your machine.

No ElevenLabs. No $99/month subscription. No per-minute pricing.

Here's what this thing does:

→ Text-to-speech that sounds indistinguishable from a real human
→ Generate up to 90 minutes of audio in a single pass
→ 4 distinct speakers in one conversation with natural turn-taking
→ Clone any voice from just 10 seconds of audio
→ Real-time streaming TTS. First audio in ~200 milliseconds.
→ Speech-to-text that processes 60 minutes of audio in one pass
→ Identifies who said what and when. Speaker labels + timestamps.
→ Supports 50+ languages for transcription
→ Custom hotwords for names, technical terms, domain-specific accuracy

Here's the wildest part:

Give it a podcast script. It generates a full multi-speaker conversation that sounds like two real humans talking. Natural pauses. Emotional nuance. Turn-taking. 90 minutes. One command.

Microsoft had to take this repo down once because people were misusing it for deepfakes and disinformation. They brought it back with embedded watermarks, audio disclaimers, and safety controls.

That's how powerful this is. A $3 trillion company built it. Released it. Pulled it. Fixed it. And gave it back to the world.

ElevenLabs: $99/month.
Play.ht: $39/month.
Amazon Polly: pay per character.

This: Free. Local. MIT License.

23.5K GitHub stars. 2.6K forks. Backed by Microsoft Research.

100% Open Source.
♥ 2.6K · ⟲ 395 · 👁 549.3KView on X ↗

Steven Kuhn Posts Political Attack on JD Vance and Peter Thiel

Steven Kuhn Posts Political Attack on JD Vance and Peter Thiel▶

Steven Eugene Kuhn posts a political message calling JD Vance a dangerous operator and suggesting he is controlled by backers, linking to the Take America Back movement site. The post offers no substantiated evidence.

Original post · 1 min read
JD Vance is the most disciplined, calculated, and patiently dangerous political operator in America right now.

The scariest part is that somebody built him.

Somebody owns him…

and when that bill comes due…

what will happen.

JoinTAB.us

29 March 2026

#JDVance #TAB #PeterThiel #LeadWhereYouLive @joerogan @patrickbetdavid @ShawnRyanShow @RealCandaceO @TuckerCarlson
jointab.usTake America Back | Citizen Servant Leaders MovementBuilding functional alternatives through the Citizen Servant Leaders doctrine. Restoring accountability at the individual level.
♥ 20.7K · ⟲ 5.5K · 👁 185.1KView on X ↗