Todd Hanford

I've been complaining about this too! It's gotten really bad recently. I've had to report dozens of email domains as phishing, but @gmail doesn't automatically block or send these emails to spam even after I report them. I've had to set up automatic rules to delete these emails because I was getting dozens a day. These emails are so easy to identify as spam!

I'm pretty disappointed how Gmail keeps assisting for email scams by allowing sender impersonation and doing absolutely nothing about it. Here is one example. This email was NOT sent by SendGrid and is a phising email, meant to take over your SendGrid account. Gmail lets it thru

The new Google "Modern Web Guidance" is a joke. The user experience "mega skill" is a single skill file with: - 8 guides all with broken links - 3 of the 8 guides are related to scrollbars... These are the most random UX guides ever. I don't even think you could prompt Claude or Codex to write a skill this bad. developer.chrome.com/docs/modern-we…

In my experience it's a "tragedy of the AI commons". Our competitors pitch our customers that they have fantastic new AI capabilities (real or not doesn't matter) and so we have to do the same. Internally, it's hard to tell whether using tools for coding and other activities provides an advantage. IMO, a big difference between a staff engineer and a junior engineer is that a staff engineer has had to live with the consequences of their coding decisions for over a year. It's too early to say whether companies are truly accelerating their engineering or they are creating a "slop bomb" that will bite them much later.

I've yet to see, or hear of a company that is winning against its competition, because said company is spending more on AI tools, or using it better than the competition. Ways I see companies win: - better product - better marketing - cheaper prices - better unit economic - more funds raised etc

@jaezun_ Looks great! Do you have to create your own chief of staff agent to create other agents or is that custom built / bootstrapped?

Day 5: introducing agent birthing find me a better agent creation UX and I will buy you coffee

For the haters in the comments, you can just one shot the agent swarm using Wispr Flow to pipe into Claude mobile connected to your Mac mini through Claude remote control. You could even run a local hosted open source model if you're concerned about cost. Just connect that to your pi agent and add tools for image parsing and generation.

@chamath User error All you have to do is setup an agent swarm with: consolidator agent splitter agent analysis subagents deck generator agent

Use Claude they said. Upload your decks the said. Unleash all this productivity they said. But apparently, I first need to start a new chat, delete some of the deck and not exceed the maximum image count…just like my existing brain.

@chamath Or have @jason setup an OpenClaw agent to run this analysis overnight for $400 in API costs. Probably best to add your credit card credentials in case your Claw needs to use external API services

@zackbshapiro 50.9% of statistics are false

Friendly reminder to would-be SaaS investors that this is not necessarily a bullish metric. It could be that monthly users are converting into daily users (as Winston seems to imply), but this metric could just as easily mean that non-daily users are dropping the tool altogether. (Or, more likely, could be some of both)

Backend + Frontend Software Engineer = Full stack developer Product manager + Designer + Backend software + Frontend software = ???? MEGA stack developer?

Not justifying the price of Harvey and Legora, but I think most people miss the point that all OpenAI / Anthropic customers don't benefit from collective learning and improvements. Harvey and Legora have thousands of customers and users which they use to optimize the product. Anyone who downloads the Claude for Legal skills can likely expect those skills will not be updated again (and haven't received anywhere near the level of scrutiny as the features built by Harvey and Legora). As a corollary, Anthropic launched Claude for Healthcare in January 2026 and their webpage hasn't been updated since (I checked the wayback machine).

I feel like I was quite early to the "How are Harvey and Legora any different from ChatGPT?" (I've got receipts) But, I now find myself being talked out of this line of thinking by the "Everything is a wrapper" crowd here on X.

Navy SEALs: "slow is fast and fast is smooth" AI pilled CEOs: "fast is fast and fast is..."

Totally get the readability and rich visualization of html files. But, as your FAQ points out, HTML is less token efficient and take 2-4x longer to generate. This seems like an affordance for humans rather than agents. I'm curious whether you ran any evals to compare agent performance when using html vs md. Seems like an easy comparison to make since the text content could be exactly the same. Headless applications have been trending up with the latest Salesforce announcement being the most notable. This is the exact opposite bet - agents don't need a UI.

@kylejeong Profitable on day 1 💪 My wife buys stuff from Amazon and then returns it for pure profit. We should collaborate.

i just started the fastest company to $ 5M ARR. sold a pair of airpods to my friend for 100 dollars, transaction took ~10 minutes from procurement to close, meaning i've hit approximately a $5,256,000 annualized run rate. taking only tier 1 investors in my dms

This is why I engineer AI applications to be flexible. Gemini 3.1 Flash-Lite Preview will deprecate on 5/24. One click to migrate to a new model. The alternative is reacting to production failures when your model stops working.

My wife is closing out our 401k to buy a Mahjong set. It's crazy how expensive a board and some tiles are! Why can't we just all go back to pickleball??

@GergelyOrosz Datadog had an outage but it was not 8 hours. They did end up updating the notification to say it was an AWS issue.

Outside of Coinbase, did any other major service have a 8-hour outage? I’ll be honest: did not notice anything else. Want to make sure I didn’t miss anything? (AWS had an outage in a single AZ. This should have… NOT taken down any service with resiliency 101)

AGENTS.md file -> prefer Pydantic models instead of dicts pre-commit skill -> check for dictionaries that would be better typed as Pydantic models Codex 🫡 def func() -> dict[str, Any] Am I dumb or does Codex hate type safety? @sama

@Hesamation They use Workday for their applicant tracking system. If you've ever tried applying for a job using Workday, you know it's better to not apply at all. So effectively Benioff is correct.

Salesforce CEO: “we’re not hiring engineers in 2026.” meanwhile their website has 126 open positions for software engineering. tech job market is such a joke.

I love the idea of applying @karpathy autoresearch to other agentic improvement solutions. Allowing agents to run in probabilistic loops (introducing variance) with an eval harness and an explicit improvement goal is one of the highest leverage uses for LLMs. I think most people treat evals as an annoying form of regression testing. But regression testing only catches bugs, it doesn't identify improvement opportunities. Once people start to understand that evals allow you to auto-improve, I think they will become much more popular.