Building and marketing projects with AI in public. Agency founder after exit, ex-VP ENG at TIER. Dad of two. Optimizing health, sleep, and training.
Budapest, Hungary
Joined May 2009
- Tweets5.7K
- Following846
- Followers2.2K
- Likes20.9K
Pinned Tweet
Past week in AI news 🧵
- Claude Duet, Anthropic's proactive agent, is now available via a waitlist.
- Dessa, Block's full-duplex speech model was released, trained purely on audio of human conversation.
- OpenAI models are coming to Google Cloud.
- Meta Muse hits 10M downloads in the US in 40 days.
- DeepSeek raised $15 Billion.
- Canva releases Canva Craft, the company's first image generation model.
- NVidia launches two new models: Nemotron 3.5 Super VL (model for always-on agents), Nemotron 3.5 Bolt (agentic intelligence to local NVIDIA systems).
- OpenAI acquired AI-powered drug development platform Orchestra Bio.
- Two new Kimi models were introduced: Kimi K3.1 Pro and Kimi K3.1 Pro Highspeed.
- OpenAI's Decisions API is now available, to help applications and agents make near real-time decisions faster.
This is not an exhaustive list, just the stories that caught my attention. Let me know if I missed anything!
What is a neocloud?
A neocloud is an AI-first cloud platform, designed to provide infrastructure for GPU compute and AI workloads.
Unlike traditional cloud platforms offering a wide range of services like databases, auth, analytics etc, neoclouds specialize in GPU compute.
Consolidation or burst of the bubble?
Pretty much.
As of the past few weeks, debt markets have almost entirely shut off access to money for any company that doesn’t have substantial revenue independent of AI data centers.
SpaceX closed $40B of debt for GPUs, but we’re investment grade and have Starlink & launch revenue.
That means the 200+ neoclouds are going to see guppy-level death and consolidation. Some will make it through, but not many.
If one of your biz strategies is “user forgets to cancel”, I’m just going to stop reading further.
This quoted post is unavailable.
This is the world we live in now
BREAKING: OpenAI says it found a critical flaw in Brotli, a widely used compression library built into software like Node.js, Bun and Envoy.
The bug could let an attacker take over a system just by sending it malicious compressed data, with no login needed. That works both ways: a hacker could hijack an unpatched server, or a malicious server could hijack an unpatched user’s device.
This is an excellent point. A (good) SW engineer has to learn continuously until the end of their career. And a great SW engineer actually enjoys learning, and teaches others as well!
I think many of us working in tech tend to forget that the reason software engineering vs other engineering fields pays (a lot!) more, is a lot more flexible (WFH/remote), and has better perks:
Because this field changes much faster than other engineering fields. The premium is because fewer people are able/willing to pick up new technologies, methodologies so quickly.
But the field had ~10-20 years of stability, and so lots of folks forgot this is why a premium exists. And that AI is just another tech wave: software engineers who pick it up and master it will generally fare better; those not will struggle a lot more.
Not in the mood for marketing but still want to tick the box?
Drop your product URL here, just the URL.
It counts.
Ever since using AI heavily for coding, I feel much more tired at the end of the day.
Is it just me?
My prediction: using AI is going to be much much more expensive soon
A new Anthropic S-1 leak has dropped. 👀
According to a draft seen by the Wall Street Journal, Anthropic recorded an $80.7 billion net loss for the first half of this year. Last year, it saw a net loss of $13 billion during the same period.
The company continues to signal it's going to spend a lot more on compute. Anthropic's computing agreements will cost around $450 billion over the next seven to ten years.
Is Opus 5.5 really nerfed?
I read this a few times on X, but can't find any evidence, and I also do not experience it being nerfed.
Build in public has to adapt.
With AI, execution isn't the bottleneck anymore, at least on the software side. Revealing what you're building can result in competitors implementing your ideas quicker, so drawbacks may outweigh benefits.
This changes what is worth sharing.
I continue to be amazed about how good Opus 5.5 is
It is great at following specs, points out potential bugs and blind spots, and the response/explanation is easy to digest
I still review every piece of code shipped to production, but this might be me being overly cautious
That is something I don't want to let go of. Not sure why, but typing git commands into my terminal is just soo satisfying!