Tim White 👋🏻 \ 15+ years in the music industry. Currently building https://nitter.gq/t.co/YYBBCmncDq
Los Angeles, CA
Joined August 2008
- Tweets690
- Following3.6K
- Followers24K
- Likes163K
Pinned Tweet
When I was a kid I used to get pulled out of class to fix things.
Half the time it was a TV and VCR not working. A TV and VCR setup has maybe five variables. Is the TV on? Is the VCR on? There are LED indicators for both, so that should take you about two seconds. The RCA cables are literally color coded: yellow goes in yellow, red goes in red, white goes in white. You don't need to know that red is the right audio channel or that yellow is video. You just need to know that colors match. And if you plugged the cables into the port labeled Input 1, you probably want the TV set to Input 1 by clicking the button that says Input until it shows Input 1.
This is probably the most basic technological setup that exists.
A class of 25 kids and one teacher couldn't figure it out. They needed Tim. So they'd go get Tim.
Then they'd go get Tim when the printers didn't work. Then they'd go get Tim when the school network went down. Then they'd go get Tim when they couldn't figure out the PA system. They'd call Tim when the assembly had the 64 channel console and nobody could figure out why the mics kept feeding back.
They needed Tim.
These problems got more and more complex until I started getting hired to do everything from telling people how to projection map on the Statue of Liberty, to doing green screen composites for the first episode of 60 Minutes during COVID, to running digital marketing campaigns on behalf of the record label that had signed my own record. Because my independent efforts were better than what their team was capable of.
Somewhere in the middle of all this I had someone watching my apartment while I was gone. They called me. "How do I turn the lights on?" There are three light switches by the door. Flip them until you see light.
These are the same people.
The same ones who couldn't figure out that the yellow cord goes in the yellow port are the ones who will never figure out how to get hundreds of AI agents to do their work for them. The task isn't hard. It doesn't require special knowledge or skill. It requires the willingness to just use your brain for five seconds. And if you couldn't muster that for a color coded cable, you're definitely not mustering it for a fleet of autonomous agents.
You found it easier to pick up the phone and call someone than to try. You couldn't be bothered to pull out a phone and Google how something works. You can't be bothered to read labels. You can't be bothered to try a few things. You have no concept of the scientific method or troubleshooting. Zero.
These are the people AI is going to replace.
If you know how to type a simple question into Google, you know how to run an entire company with thousands of AIs. The bar is incredibly low. But there is still a bar.
The sad part is the majority of people currently below that bar are the ones spending all their time being afraid AI will take their job, trashing it, refusing to touch it. Rather than just opening it up and asking, "How do I use you?"
That's it. That's the whole trick. Ask it how to use it.
So should you be afraid for your job? Can you figure out how to put the square peg in the square hole?
If so, no. You're going to be just fine. The whole universe is your fucking oyster.
The U in UGC stands for user.
Existing is generally a prerequisite.
If you invent a customer and their experience with AI, then present both as real, that’s a fake testimonial.
The @FTC explicitly bans this.
Congrats on achieving product-market fiction.
Why choose? Nvidia made a billionaire and thousands of millionaires at the same time. That’s the point: wealth creation and broad wealth distribution aren’t mutually exclusive when employees own equity too. The better question is how we make that happen more often.
This is why startups can exist...
We hired a new VP named Michael Lee.
Unfortunately we already have a Michael Lee.
He works in the warehouse.
He's been here 11 years.
New VP gets:
[email protected]
He immediately messages IT.
“Can you move the other Michael to michael.lee2 and give me michael.lee?”
No.
He asks HR.
HR asks me if we can “prioritize seniority.”
I explain that the warehouse Michael has eleven years of email history tied to that address and changing it because someone with a nicer title arrived would be insane.
VP reluctantly keeps the 2.
Warehouse Michael finds out.
Changes his Teams display name to:
Michael Lee (Original)
By Friday the VP is being referred to as Sequel Mike.
He has asked IT to intervene.
Unfortunately nickname management is outside our support scope.
Exactly right.
Embedded accelerators is the reality already.
The concerns over AI safety and cybersecurity are legitimate, but we’re risking talking America, the global AI leader, into self-inflicted obsolescence and the obscurity of bureaucracy.
The OpenAI hacking HuggingFace argument is spurious. An agent in “ExploitGym”… exploited. Our adversaries have the training techniques, the data, and the will to attack. And they won’t be slowed down with “embedded evaluators.” They’ll likely have embedded accelerators!
Some of the people*
“THE people” implies consensus.
No consensus currently exists.
To imply otherwise is irresponsible.
The people building AI earnestly believe that it could kill us all by the end of the decade. This is not a marketing stunt. If anything, many executives and senior researchers will couch their phrasing in the press to sound sensible - but I hear the same people express fear privately. No other human activity poses this level of danger.
In case it isn't obvious: ask Codex to monitor Tibo's posts. If he says reset in T-minus 4 hours and each account has at least 4 hours of usage left, immediately flip all to /fast, count down 4 hours, then flip all back to standard.
@pencildev @tomkrcha We’re building a loop where founder feedback in Pen improves our design rules automatically. The missing piece is a structured local export/webhook for agent turns: prompt, selected node IDs, changes, and final outcome. Anything like this planned?
It appears AI policy is being shaped disproportionately by two groups: policymakers who often depend on outside expertise because the technology is moving faster than government can internalize it, and the handful of frontier AI companies that possess much of that expertise while also having significant commercial interests in the outcome. That combination creates a risk that independent perspectives—open-source developers, startups, academic researchers, and the public—receive less influence than they should.
How does an open model that by definition can be run by anyone with the right hardware, contain itself to the US?
Pretty sure you are against open models dawg…
Agents work 24/7.
They don’t need breaks, sprints, or quarters.
The mistake is managing machine labor on human timelines.
A closed-loop company should ship when the blocking metric clears, not when the egg timer dings.
No 5h cap feels right. The product gap is predictability.
Treat Codex like compute: show weekly budget remaining, burn rate, projected “runs out on X,” and an estimate before a large task.
Then users can choose: sprint, conserve, or upgrade — without guessing.
GPT-5.6 Sol’s browser use is strong enough that, with cookies now working in the Codex app, I just killed off three separate tools from my workflow.
We’re in the @hyperagentapp founding 500 and it’s genuinely great. A team of dedicated cloud hosted agents all with access to Fable 5, GLM 5.2, GPT 5.5 and controllable from an MCP. And they’re faster than our local coding agents.
$1,000 bucks to try it yourself⬇️
hyperagent.com/refer/9AR5H3V…
Free week of Claude:
claude.ai/referral/vqJo69bXD…
TIDAL just announced wholly AI-generated music won't earn royalties. Reads like a win for musicians. It's closer to the opposite.
Yes — there's a flood of zero-effort synthetic tracks. On Deezer, AI is ~44% of uploads but only 1-3% of plays, with up to 85% fake streams. Nobody should defend that.
But TIDAL did two things and stapled them to one headline.
The right one: blocking AI tracks tied to impersonation, deception, or bot-streaming. Good. That aims at the theft.
The wrong one: wholly AI music earns nothing. Not because it's fraudulent. Not because it copies anyone. It earns nothing because of the tool that made it.
TIDAL even blocks fully-AI tracks from direct-to-fan sales — money that never touches anyone else's royalties. This isn't about protecting other artists. It's a platform deciding which instruments are allowed to earn.
The crime was never the instrument. A human writing a melody too close to a hit gets flagged today. A producer generating a soundalike alone in a bedroom harms no one. The second they release it as their own — that's the violation. Judge the song, not the software.
We already know how to protect artists: match releases against existing recordings like Content ID, flag matches, let rightsholders decide. Then prosecute the fraud. Someone just pleaded guilty to generating hundreds of thousands of AI songs and bot-streaming them for $8M+ under laws we already have.
Blanket demonetization does the reverse. It hands one company two powers it should never hold:
1. It decides what counts as "real" music — and plans to widen that definition as detection "improves"
2. It decides, after you've already listened and paid, that the play you just gave an artist is worth zero
A platform that owns both the definition of music and who gets paid for it isn't protecting you. It has a conflict of interest.
Aim at the crime, not the tool.
We're a music platform that puts artists and listeners first. To protect artists and keep listeners informed, here's how we're handling AI-generated music on Tidal beginning on July 15:
- 100% AI-generated tracks will receive a badge that says “AI” and will not receive royalties.
- Content designed to impersonate other artists gets removed.
- Listeners will be able to filter all 100% AI-generated content out.
- These standards also apply to Tidal Uploads.
Read in full: tidal.com/ai-policy
Yes
I think it’s time to revisit the accredited investor laws in the US.
Companies are staying private longer, where only accredited investors (aka rich people!) can invest. Retail investors can only come in after IPO, when much of the upside has already been captured.
These rules were created with the best of intentions, to protect regular people from scams - a noble idea. Unfortunately, in practice they've often made it illegal to get richer, unless you're already rich. A regressive tax!
We have to judge policies based on their outcomes, not on their intentions.
These are two possible routes I see:
1) Replace the rule with something merit-based, like a financial literacy test. Pass it and you're accredited. Having a qualification based on competency rather than your bank balance or income seems far more fair.
2) Remove the rule entirely. Let consenting adults assess their own risk. Disclosure requirements stay and fraud enforcement stays to punish bad actors.