OpenAI Buys Astral + Bezos' $100B AI Fund

OpenAI is acquiring Astrol, a startup that builds open source tools for developers.

Astrol's team will join OpenAI's Codex group, which runs their AI coding assistant.

This is OpenAI's fourth acquisition in less than a year following deals in cybersecurity,

healthcare, and AI devices.

Codex now has over 2 million weekly active users, and that number's tripled since the start

of the year.

I think this signals how intense the competition is getting with Anthropic in cursor.

Yeah, and OpenAI hired Google's Albert Lee to lead corporate development back in December,

so expect more of these acquisitions.

They're clearly in aggressive expansion mode, buying their way into key developer tooling.

Makes sense, if you can't build it fast enough, by the team that already has traction

with developers.

Deaf Bezos is raising $100 billion fund to buy manufacturing companies and automate them

with AI.

He's been meeting with sovereign wealth funds in the Middle East and asset managers in

Singapore to secure backing.

$100 billion, that's like one of the largest industrial buy-up vehicles ever.

The idea is to target traditional manufacturers that are way behind on automation and basically

apply Amazon's playbook at industrial scale.

Right, and if this works, we're talking about a massive wave of AI-led consolidation

in manufacturing.

I'm curious how private equity firms react, because they're probably building similar

strategies.

Yeah, this could get competitive fast, but Bezos has the track record and the relationships

to actually pull this off.

Harris-based parallel just raised $20 million in Series A funding to deploy AI agents

that automate hospital administrative work.

Specifically, they're targeting medical coding, where specialists spend entire days converting

patient records into standardized billing codes.

Every patient discharge creates this paperwork cascade.

You've got to pull records, navigate legacy systems, enter the right ICD codes, it's

painstaking and manual, and parallels AI agents are doing it end-to-end.

The Rapid Series A timeline suggests they've got real traction, and honestly, with staffing

shortages in healthcare, automating high-volume clerical work isn't a future play, it's an

immediate need.

Absolutely, this directly impacts hospital margins and frees up staff capacity.

I think we'll see a lot more AI agents tackling these kinds of administrative bottlenecks.

Oasis security raised $120 million led by craft ventures with Sequoia, Axel, and Cyber

Starts participating.

They've now raised $190 million total to build tools that manage access from non-human

accounts like AI agents.

This is a growing security gap, as companies deploy AI agents that need system access, but

don't have traditional authentication methods, managing these non-human identities becomes

critical.

Yeah, and the size of this round tells you investors think AI agent security will be

a major enterprise category.

As autonomous AI systems interact with corporate infrastructure at scale, you need a whole

new security layer.

Definitely.

Expect more startups and incumbents to jump into the space.

It's going to be a big market.

Starting next month, US Fitbit users can link their medical records to the app's AI

health coach.

We're talking lab results, medications, visit history, all feeding into personalized advice

based on your wearable data, too.

So instead of generic answers about cholesterol, you can ask how to improve yours and get advice

based on your actual medical history.

Google says the data won't be used for ads, but there's a disclaimer that Fitbit can't

diagnose or treat condition.

This follows similar moves by Amazon, OpenAI, and Microsoft.

They're all betting users will trade sensitive health data for personalized AI coaching.

But companies face serious regulatory scrutiny from the FDA, and these products aren't even

available in Europe yet due to privacy laws.

Yeah, there's a fine line between wellness coaching and medical advice.

I think people should be careful about what they share, especially reproductive data in

the US.

That's today's Fitbit 5.

Like and share Fitbit 5.

Thanks for listening.