Nullshot

One-click AI agents for non-technical humans. They do the real work, you stay in control.

[Nullshot @nullshotai]

https://t.co/hhf5CTL2TN

[Nullshot @nullshotai]

Anthropic released Claude Fable 5.2 and Mythos 5.2 this week, and the examples are wild.

Fable 5.2 built a working operating system from an empty folder, designed three electronic devices that worked first time, and helped formalize Fermat’s Last Theorem in 30 hours. It’s also up to 30% faster than 5.1, and cheaper.

Both are built on the same model: Fable is open to everyone, while Mythos is available to verified cybersecurity and life sciences organizations.

The most useful detail is easy to miss: for everyday work, Anthropic still recommends Opus 5.5. The most powerful model isn’t always the right one for the job.

[Nullshot @nullshotai]

New Claude models just dropped. Coding, reasoning and vision all get a bump.

[Claude @claudeai]

Introducing Claude Fable 5.2 and Claude Mythos 5.2, designed to tackle the most daunting technical challenges.

They’re capable collaborators on ambitious projects in many fields, and show improvements in areas like coding, reasoning, and vision.

[Nullshot @nullshotai]

Who even replies to their own emails anymore?

[Nullshot @nullshotai]

Top AI image models right now:

- GPT Image 2.5 Sunburst (OpenAI)
- Nano Banana 2.1 (Google)
- Grok Imagine Image 2.0 (xAI)
- MAI-Image-2.6 (Microsoft)
- Muse Image (Meta)

[Nullshot @nullshotai]

2030: I just want to be the human in the loop.

[Hardik Gohil @GohilHardy]

2018: I just want to become a software engineer.

2026: I just want to remain a software engineer.

[Nullshot @nullshotai]

Slow Saturday. Letting the agents handle the boring stuff.

[Nullshot @nullshotai]

If it felt like a new AI model came out every other day in October, that’s because one did.

[Nullshot @nullshotai]

Declining gracefully

[Nullshot @nullshotai]

If you’ve got an agent set up but haven’t really used it yet, the weekend’s a good time to think about what you do every day that’s a bit repetitive.

Checking the same few sites each morning, copying info from one app to another, that kind of thing. Usually there’s at least one you could hand over.

[Nullshot @nullshotai]

“You are an expert” at the start of a prompt is the “Dear Sir or madam” of 2026

[Nullshot @nullshotai]

Scorpion stings kill more than 3,250 people a year, mostly children, and there’s little money in new treatments. Claude spent 64 hours designing proteins to bind the toxins, and 17 bound their target in lab tests.

[Anthropic @AnthropicAI]

New on our Science Blog: designing protein binders against scorpion neurotoxins.

Scorpions sting about 1.2m people every year, and antivenoms are often costly or in short supply. Claude has made an early step toward finding new treatments.

https://t.co/l6LgdEqM7I

[Nullshot @nullshotai]

Worth the wait

[Nullshot @nullshotai]

A few people on the team have sent me the same article about Meta’s Muse Wonder today. It remembers what matters to you across conversations, and when a task gets hard, it keeps going, tries another way and recovers from mistakes.

Is this a hint?

[Meta Newsroom @MetaNewsroom]

Meta is introducing Muse Wonder and Muse Wonder Full, our most capable models to date, built to deliver state-of-the-art performance on personal and professional agentic tasks.

https://t.co/AOgIF0ThUW

[Nullshot @nullshotai]

Nobody talks about how hard it is to name your agent. I’m going to be talking to this guy every day, this is more serious than naming a pet

[Nullshot @nullshotai]

Peak productivity

[Nullshot @nullshotai]

I just realised I make most of my buying decisions based on what an AI tells me now. And it's just repeating what it read on Reddit.

[Nullshot @nullshotai]

The 3 growth channels founders sleep on (and why):

1. Replying, not posting.
Everyone grinds on their own feed shouting into the void. The real reach is in the replies under bigger accounts in your space. Borrowed audience beats no audience. Founders skip it because it feels less like "their" content.

2. AI answers.
When someone asks ChatGPT "best tool for X," something gets recommended. That's the new search bar. Founders sleep on it because it's invisible; you can't see the traffic, so nobody optimizes for it.

3. Communities where people ask for recommendations.
The highest-intent buyers alive are the ones literally typing "what should I use for this." Founders skip these places because they're scared of coming off as spammy, so they leave free, ready-to-buy demand on the table.

[Nullshot @nullshotai]

Nobody's short on content. Everyone's short on trust.

[Nullshot @nullshotai]

I just realised I don't trust reviews on a company's own site at all. But 4 people agreeing in a Reddit thread? Sold.

[Nullshot @nullshotai]

How to write a Reddit comment people actually upvote:

- answer the question in the first line
- be honest, even about where your product falls short
- skip the link
- sound like a person, not a brand