Tech
‘Not built right the first time’ — Musk’s xAI is starting over again, again
And then there were two: Of the original 11 co-founders who kickstarted xAI with Elon Musk three years ago, only two remain as the deep learning lab continues a personnel overhaul to compete with Anthropic and OpenAI. That rebuilding, insists Musk, is by design.
“xAI was not built right first time around, so is being rebuilt from the foundations up,” Musk said Thursday on his social media platform, X. By most measures, it isn’t going all that smoothly.
The most immediate pressure is competitive. This week, xAI co-founders Zihang Dai and Guodong Zhang left the outfit after Musk complained that the company’s AI coding tools were not effectively competing with Claude Code or Codex, rival programming assistants made by Anthropic and OpenAI, respectively. Musk said the company held an all-hands meeting on Wednesday that focused on how to catch up, which he predicted would be possible by the middle of this year.
Coding tools matter so much because they’re where the money is. While an early-year surge of users was powered by xAI’s lax regulation of Grok’s ability to produce sexual and even abusive imagery, coding tools are seen as the key revenue-generating tech for AI labs. That makes xAI’s current lag in this area more than a perception issue; it’s a business problem.
The personnel overhaul extends well beyond this week. A month ago, 11 senior engineers at xAI, including two co-founders, left the company following changes Musk described as a reorganization to suit a larger business. That effort was apparently insufficient: The Financial Times reported that SpaceX and Tesla executives have parachuted into the company to evaluate employees and fire those who don’t make the grade.
The two remaining co-founders, Manuel Kroiss and Ross Nordeen, along with Musk, have their work cut out for them.
Musk is now casting a wider net for talent. On Thursday, he said on X that he and another colleage, Baris Akis, are currently reviewing rejected employment applications in the company, with an eye toward reaching out to promising candidates who should have had a chance to interview. “My apologies,” Musk added, addressing the pile of strangers he’d ghosted.
Techcrunch event
San Francisco, CA
|
October 13-15, 2026
For the sake of comparison, LinkedIn reports that xAI has just over 5,000 employees, compared to more than 7,500 at OpenAI and more than 4,700 at Anthropic.
On the hiring front, there’s at least one encouraging sign. Andrew Milich and Jason Ginsberg are joining xAI from the AI coding tool company Cursor, where the two held joint responsibility for product engineering. Unlike xAI, Cursor depends on frontier labs for access to the AI models it runs on. Their decision to join xAI may signal the importance of direct access to LLM and computing resources to run them — and suggest that xAI’s core asset, its own frontier model, is still an attractive draw.
Either way, the pressure to show results is as much external as it is internal. Now that xAI is part of SpaceX, and with a public offering of SpaceX shares anticipated, the cash-burning unit is under pressure to demonstrate real uptake on Grok, its LLM. (A stumbling AI division is not the story Musk needs investors to be reading.)
Longer term, Musk is betting on something bigger than coding tools. xAI’s Macrohard project — Musk is convinced the name is “a funny reference to Microsoft” — aims to create an AI agent capable of doing anything a white-collar worker can do on a computer. Toby Pohlen, chosen to lead the project in February, left within weeks, and this week, Business Insider reported that Macrohard was on pause.
Musk’s response has been to draft another of his companies into the project. He revealed for the first time that Macrohard is a joint effort with Tesla, which is also developing a complementary agent dubbed “Digital Optimus” — a reference to Tesla’s Optimus humanoid robot. In Musk’s description, the xAI language model would direct the Tesla agent as it performs tasks.
It’s ambitious; it’s also not unique. Instead, the vision is not far off from what Perplexity — an AI-powered search engine — is doing with its new “Everything is Computer” offering, which aims to offer enterprise users a dedicated “digital proxy” that can orchestrate their digital tasks. It also echoes what entrepreneur Peter Steinberger is now working on at OpenAI, after creating OpenClaw’s popular personal agents.
Tech
Spotify will let you edit your Taste Profile to control your recommendations
At the SXSW conference on Friday, Spotify co-CEO Gustav Söderström announced a new feature, launching in beta, that will allow listeners for the first time to review and edit their Taste Profile, the algorithmically generated model of their music preferences.
This Taste Profile is key to Spotify’s recommendations, including personalized playlists like Discover Weekly, Made For You recommendations, and the year-end review known as Spotify Wrapped, among other things.
Starting with Premium listeners initially in New Zealand, Spotify will allow users to see all their listening data in one place in the app, including music, podcasts, and audiobooks. Users will then be able to edit this profile and even fine-tune future recommendations by asking for more or less of a certain vibe. After doing so, the app’s home page will reflect a different set of suggestions.

To access the Taste Profile, users tap on their profile pic, then scroll down. Changes can be made using natural language prompts.
Spotify had previously offered some tools to remove music from your Taste Profile before, but they were not as comprehensive. Instead, users were only able to exclude certain tracks or playlists from their profile. Because of this, and the largely hidden nature of the Taste Profile overall, Spotify users often complained that the app’s recommendations didn’t reflect their interests.

Today, users often share their Spotify account with others, like family members who access their account through a shared smart speaker or smart TV in the living room, for example, or teens who take over in CarPlay while they drive.
Other times, users may listen to music that they don’t want to characterize as their “taste,” like the sleep sounds or quiet tracks they play at night, or music to entertain their kids. Users don’t always remember which tracks or playlists need to be removed, nor do they have time to go back and do so. This can lead to the Taste Profile becoming cluttered with music users don’t like.
Techcrunch event
San Francisco, CA
|
October 13-15, 2026
It also significantly impacted, even ruined, many people’s annual Wrapped experience in the app — particularly because of kids’ use of their parents’ Spotify accounts. For years, Spotify users have asked for a fix for this problem.
Spotify says the Taste Profile feature will roll out in the coming weeks in New Zealand before expanding to other markets.
Tech
The wild six weeks for NanoClaw’s creator that led to a deal with Docker
It’s been a whirlwind for NanoClaw creator Gavriel Cohen.
About six weeks ago, he introduced NanoClaw on Hacker News as a tiny, open source, secure alternative to the AI agent-building sensation OpenClaw, after he built it in a weekend coding binge. That post went viral.
“I sat down on the couch in my sweatpants,” Cohen told TechCrunch, “and just basically melted into [it] the whole weekend, probably almost 48 hours straight.”
About three weeks ago, an X post praising NanoClaw from famed AI researcher Andrej Karpathy went viral.
About a week ago, Cohen closed down his AI marketing startup to focus full-time on NanoClaw and launch a company around it called NanoCo. The attention from Hacker News and Karpathy had translated into 22,000 stars on GitHub, 4,600 forks (people building new versions off the project), and over 50 contributors. He’s already added hundreds of updates to his project with hundreds more in the queue.
Now, on Friday, Cohen announced a deal with Docker — the company that essentially invented the container technology NanoClaw is built on, and counts millions of developers and nearly 80,000 enterprise customers — to integrate Docker Sandboxes into NanoClaw.
Scary security of OpenClaw
It all started when Cohen launched an AI marketing startup with his brother, Lazer Cohen, a few months ago. The startup offered marketing services like market research, go-to-market analysis, and blog posts through a small team of people using AI agents.
Techcrunch event
San Francisco, CA
|
October 13-15, 2026
The agency started booking customers, and was on track to hit $1 million in annual recurring revenue, the brothers told TechCrunch.
“It was going really well, great traction. I’m a huge believer in that business model of AI-native service companies that have margins and operate like a software company but are actually providing services,” said Cohen, a computer programmer who previously worked for website hosting company Wix.
He had built the agents the startup was using, largely using Claude Code, each designed to do specific tasks. But there was “a piece” missing, he said. The agent could do work when prompted, but the humans couldn’t pre-schedule work, or connect agents to team communication tools like WhatsApp and assign tasks that way. (WhatsApp is to most of the world what Slack is to corporate America.)
Cohen heard about OpenClaw, the popular AI agent tool whose creator now works for OpenAI. Cohen used it to build out those final interfaces, and loved it.
“There was this big aha moment of: This is the piece that connects all of these separate workflows that I’ve been building,” he said and immediately decided, “I want more of them: on R& D, on product, on client management,” one for every task the startup had to handle.
But then OpenClaw scared the bejesus out of him.
In researching a hiccup with performance, he stumbled across a file where the OpenClaw agent had downloaded all of his WhatsApp messages and stored them in plain, unencrypted text on his computer. Not just the work-related messages it was given explicit access to, but all of them, his personal messages too.
OpenClaw has been widely panned as a “security nightmare” because of the way it accesses memory and account permissions. It is difficult to limit its access to data on a machine once it has been installed.
That issue will likely improve over time, given the project’s popularity, but Cohen had another concern: the sheer size of OpenClaw. As he researched security options for it, he saw all the packages that had been bundled into it. It included an “obscure” open source project he himself had written a few months earlier for editing PDFs using a Google image editing model. He had no idea it was there — he wasn’t even actively maintaining that project.
He realized there was no way for him to validate all OpenClaw’s code and its dependencies, which, by some estimates, sprawled across 800,000 lines of code.
So he built his own in just 500 lines of code, intended to be used for his company, and shared it. He based it on Apple’s new container tech, which creates isolated environments that prevent software from accessing any data on a machine beyond what it is explicitly authorized to use.
Going viral
At 4 a.m., a couple of weeks after sharing it on Hacker News, his phone started ringing non-stop. A friend had seen Karpathy’s post and was urging Cohen to wake up and start tweeting, which he did, setting off a public discussion with the well-known AI researcher.
Attention to NanoClaw followed like a landslide. More tweets, YouTube reviews from programmers, and news stories. A domain squatter even snagged a NanoClaw website URL. The correct one is nanoclaw.dev.
Then Oleg Šelajev, a developer who works for Docker reached out. Šelajev saw the buzz and modified NanoClaw to replace Apple’s container technology with Docker’s competing alternative, Sandboxes.
Cohen had no hesitation about pushing out support for Sandboxes as part of the main NanoClaw project. “This is no longer my own personal agent that I’m running on my Mac Mini,” he recalled thinking. “This now has a community around it. There are thousands of people using it. Yeah, I said, I’m going to move over to the standard.”
For all the changes these weeks have brought Cohen and his brother Lazer, now CEO and president of NanoCo, respectively, one area still needs to be figured out: how NanoCo will make money.
NanoClaw is free and open source and, as these things go, the Cohens vow it always will be. They know they would be strung up as villains if they ever betrayed the open source community by changing that. Currently the Cohens are living on a friends-and-family fundraising round, they said.
While they are cautious about announcing their commercial plans — in large part because they haven’t had a chance to fully formulate them — VCs are already calling, they say.
The game plan is to build a fully supported commercial product with services including so-called forward-deployed engineers — specialists embedded directly with client companies to help them build and manage their systems. This will likely focus on assisting companies in building and maintaining secure agents. That is, however, a crowded field growing more crowded by the hour.
But given the giant community of developers that NanoClaw just unlocked with Docker, we’re sure to hear more about this soon.
Pictured above from left to right, Lazer and Gavriel Cohen.
Tech
Travis Kalanick launches a new company called Atoms focused on robotics
Uber founder Travis Kalanick has a new company called Atoms focused on robotics that, according to its website, will operate in the food, mining, and transportation industries.
Kalanick is rolling his existing ghost kitchen company, CloudKitchens, into Atoms. It’s not immediately clear how he plans to tackle mining and transportation. Atoms’ website says it will build a “wheelbase for robots,” and Kalanick said in a live interview with TBPN on Friday that his company will apply this wheelbase to “specialized robots” — not humanoids.
“Humanoids have their place, but there’s a lot of room for specialized robots that do things in an efficient, sort of industrial-scale kind of way, which is sort of where we play,” he said.
To support the mining business, Kalanick said Friday that he’s on the precipice of acquiring Pronto, the autonomous vehicle startup focused on industrial and mining sites that was created by his former Uber colleague, Anthony Levandowski. Kalanick revealed Friday that he is already the “largest investor” in Pronto.
“The industrial thing is sort of like, probably, our main jam,” Kalanick told TBPN. Kalanick demurred on the idea of using Atoms robots to move people, at least in the near-term. “Once you crack movement in the physical world, there’s lots of people who want access to that.”
Earlier Friday The Information reported Kalanick was getting back into self-driving vehicles with “major backing” from Uber, and that he has reportedly told people he “wants to be more aggressive in rolling out self-driving technology than Waymo.” Uber didn’t immediately respond to a request for comment. Atoms’ website makes no mention of Uber. The Information first reported Kalanick was discussing acquiring Pronto.
Last year, Kalanick was said to be interested in buying the U.S. arm of Chinese self-driving vehicle company Pony AI with backing from Uber, though The Information said Friday that those talks ended.
Kalanick resigned from Uber in 2017 after a confluence of crises at the ride-hail company. At the time, the company was plagued by complaints of sexual harassment and discrimination, which sparked an external investigation that resulted in more than 20 employees being fired.
Before that, Kalanick had created a self-driving division at Uber in 2015. Levandowski played a big role in that project after Kalanick lured him away from Google. Uber was ultimately sued by Google for stealing secrets related to its own self-driving car project (which eventually became Waymo). The two companies settled, but Levandowski was criminally charged and sentenced to 18 months in prison for his role in the affair. The engineer received a last-minute pardon from President Trump at the end of his first term.
The company kept working on the project after Kalanick resigned, including after one of its test vehicles struck and killed a pedestrian in 2018. Kalanick’s successor, Dara Khosrowshahi, shuttered and sold the division to autonomous trucking company Aurora in 2020.
In a rare interview in March 2025, Kalanick expressed regret that Uber had abandoned developing its own self-driving cars.
This story has been updated to reflect new information from Atoms’ website and an interview with Kalanick.
