Ted Mao

7/ If you’re interested trying out the platform or hearing more about what I’ve built, let me know!

6/ And I’ve enabled third-parties to do the same, through the OAuth authenticated MCP server. I wrote a longer post about it here: photo-intelligence.beehiiv.com/p/gumnut-mcp...

5/ I can now ask Claude to organize, label, and curate my photos — exactly the way I like them. I can run background agents to automatically categorize new photos and share them with the right people.

4/ Behind the scenes, I’ve been prototyping what a photo intelligence platform could look like and do. I recently finished a milestone that I’m particularly proud of: enabling agentic workflows around photos.

3/ There’s clearly a lot of pain, especially in workflows around curation of photos and creation of digital and physical goods from photos. There’s clearly a lot of opportunity as well.

2/ Since then, I’ve been splitting my time between product discovery and building. I’ve had the privilege of speaking with many passionate and insightful people about their use cases, pain points, and ideas for products in the space.

Gumnut + MCP = Agentic Photos

1/ Earlier this year, I started working on a side project called Gumnut, around the theme of photo intelligence: the idea that AI can use photo understanding to enable businesses and people to do things that were previously considered impossible, incredibly time consuming, or too expensive.

Starliner’s flight to the space station was far wilder than most of us thought “Hey, this is a very precarious situation we’re in.”…

They downplay it in the article, but this sounds terrifying. arstechnica.com/space/2025/0...

The Reality of Tech Interviews in 2025 Interview processes are changing in a tech market that’s both cooling AND heating up at the same time. A deepdive with Hello Interview founders, Evan King and Stefan Mai

The latest article on the Pragmatic Engineer is pretty relevant if you’re a hiring manager, or a software developer searching for what’s next: newsletter.pragmaticengineer.com/p/the-realit...

Tracing the thoughts of a large language model Anthropic's latest interpretability research: a new microscope to understand Claude's internal mechanisms

Glad someone is investing in this kind of research instead of simply plowing ahead. And I suspect that the foundational research Anthropic is doing here will also help them build better models in the future. www.anthropic.com/research/tra...

Reclaiming "Affordable Housing" Fixing housing's terminology problem

www.thesisdriven.com/p/reclaiming...

Reflections on Palantir A retrospective of an eight-year stint.

Great reflection on 8 years at Palantir: nabeelqu.substack.com/p/reflection...

I’d personally heard of the FDE model before, but while reading this article, I realized that there are a lot of similarities between it and what I’ve been doing at TheGP.

The Trump Administration Accidentally Texted Me Its War Plans U.S. national-security leaders included me in a group chat about upcoming military strikes in Yemen. I didn’t think it could be real. Then the bombs started falling.

This is an incredibly fascinating look into how senior members of the Trump administration communicate and operate. www.theatlantic.com/politics/arc...

An Interview with OpenAI CEO Sam Altman About Building a Consumer Tech Company An interview with OpenAI CEO Sam Altman about building OpenAI and ChatGPT, and what it means to be an accidental consumer tech company.

Stratechery interview with Sam Altman: stratechery.com/2025/an-inte...

The Doctor, the Biohacker, and the Quest to Treat Their Long COVID One of them is a doctor who likes to play it safe. The other is a regular dude who’s willing to try anything. Both of them have long COVID. In their desperate search for answers, two men are bonded by...

Just an interesting read. www.menshealth.com/health/a6414...

Automated Driving in Winter Conditions An Interview with Steven Waslander

Interesting article on what it will take to get autonomous cars to cities with snow. www.changinglanesnewsletter.com/p/automated-...

AI coding assistant refuses to write code, tells user to learn programming instead Cursor AI tells user, “I cannot generate code for you, as that would be completing your work.”…

Some humor: arstechnica.com/ai/2025/03/a...

Chinese towers and American blocks - Works in Progress China builds towers in a park, while America, and nearly everyone else, builds squat mid-rise blocks. The difference comes down to regulation, not culture.

This one too. worksinprogress.co/issue/chines...

Philip Oldfield on X: "Important research by the @pewtrusts shows that 4-6 storey apartments with a single fire stair do not put residents at any greater fire risk (as compared to 2 stairs) Single stairs are also cheaper, and could allow much more housing to be built! https://t.co/K2iWwTPvVZ https://t.co/1FtGN3HfX7" / X Important research by the @pewtrusts shows that 4-6 storey apartments with a single fire stair do not put residents at any greater fire risk (as compared to 2 stairs) Single stairs are also cheaper, and could allow much more housing to be built! https://t.co/K2iWwTPvVZ https://t.co/1FtGN3HfX7

I always find research on city and building design interesting. For example, the article on parking spaces that I shared a couple of weeks ago. And today, I found this interesting post on fire stairwells. x.com/SustainableT...

Here’s how I use LLMs to help me write code Online discussions about using Large Language Models to help write code inevitably produce comments from developers who’s experiences have been disappointing. They often ask what they’re doing wrong—h...

Great writeup about using LLMs for coding tasks. I also agree with the takeaway: "LLMs amplify existing expertise"
simonwillison.net/2025/Mar/11/...

I outsourced my memory to an AI pin and all I got was fanfiction Never have I ever been this gaslit by a wearable.

Kind of funny:
> There’s a glimmer of a good idea here. But after a month of testing, I’ve never felt more gaslit.

www.theverge.com/reviews/6270...

A 10x Faster TypeScript - TypeScript Embarking on a native port of the existing TypeScript compiler and toolset to achieve a 10x performance speed-up.

Very excited about this, and generally the movement toward optional and static typing in web applications. devblogs.microsoft.com/typescript/t...

How I Code With LLMs These Days Phillip Carter provides guidelines around using LLMs in software development, like paying for Claude, asking for small changes, and more.

Pretty good write-up on how to effectively use LLMs for coding. I've found the part about providing instructions to guide the model (e.g. telling it what version of python I'm using, what best practices I wanted to follow, etc.) has been particularly helpful for me.
www.honeycomb.io/blog/how-i-c...

A quarter of startups in YC's current cohort have codebases that are almost entirely AI-generated | TechCrunch With the release of new AI models that are better at coding, developers are increasingly using AI to generate code. One of the newest examples is the

My instinct is to stay far away from this latest batch of YC companies... but maybe one of them will prove me wrong by shipping a quality product? techcrunch.com/2025/03/06/a...

Two Americas, one bank branch, and $50,000 cash Ever wondered what happens if you try to take $50,000 in cash out of a bank? Answer: a year of investigative journalism.

Another great article by patio11: www.bitsaboutmoney.com/archive/two-...

How to Scale Your Product from Figma's CPO Figma CPO Yuhki Yamashita offers up his playbook for the gnarly phase of product building that arrives after a startup clinches product-market fit — and sets its sights on the next phase of growth.

review.firstround.com/lessons-in-p...

Chaithanya Kumar on X: "I think everyone should watch @Karpathy's latest video on how he uses LLMs, even those who think AI is already a big part of their lives because: 1. One of the best minds in AI is spending time showing how he uses AI personally rather he could have spent his time building AGI , https://t.co/4ldmsB7Hel" / X I think everyone should watch @Karpathy's latest video on how he uses LLMs, even those who think AI is already a big part of their lives because: 1. One of the best minds in AI is spending time showing how he uses AI personally rather he could have spent his time building AGI , https://t.co/4ldmsB7Hel

x.com/ChaithanyaK4...

The Job Market Is Frozen Unemployment is low, but workers aren’t quitting and businesses aren’t hiring. What’s going on?

I know a few people here have been looking for a job for some time, and it’s sort of validating to know that they aren’t alone in their struggles.

www.theatlantic.com/ideas/archiv...

NotebookLM - Analyzing Poop & Fart written 1,000 times - Creating meaning from the meaningless YouTube video by sorryaboutyourcats

A fascinating podcast that NotebookLM generated with some… unusual source material. www.youtube.com/watch?v=ftg7...

NotebookLM - Analyzing Poop & Fart written 1,000 times - Creating meaning from the meaningless YouTube video by sorryaboutyourcats

A fascinating podcast that NotebookLM generated with some… unusual source material. www.youtube.com/watch?v=ftg7...