X / Twitter
Your agent sees the world through string tokens while most of your data warehouse and possibly your SQL DB is not textual... Is that a problem? https://t.co/VgUCWAZuTy
Working on the hono/mcp docs, still needs some polishing.
What examples do you wanna see? https://t.co/WuyAg3UkX6
Here's 13 API's you probably need to know about to get started with Effect.
Rest of it, you can JIT
---
Basics
Effect.gen (declare a function where you can yield* other effects inside)
Effect.fn (like Effect.gen, but can receive custom parameters)
.pipe (add modifiers to an Effect)
Effect.runPromise (run your effect as a promise)
Effect.all (Promise.all, but for effects)
Errors
Effect.try / Effect.tryPromise (wrap non-Effect code)
Effect.mapError (map the errors in non-Effect code to your custom error inside a .pipe)
Data.TaggedError (create a custom error)
Effect.catchTag (catch an error)
Dependency Injection
Effect.Service (declare an interface that can be later injected)
Effect.provide (provide an implementation of an Effect.Service to a .pipe)
FileSystem.FileSystem (inject a file system implementation later for testing)
you can just position: sticky things
basic CSS layout tricks get overlooked so much, but often provide greater impact, going unnoticed
other way to trick sticky positioning 👇🧵 https://t.co/uL7Szjj9T4
Use Effect AI to embed chunks of the novel Dracula.
Store those chunks in a Postgres database using pgvector and the Effect @DrizzleORM integration.
Do a similarity search. https://t.co/6tYqPjc0Bz
Keith Rabois: “The velocity of your company improves by adding barrels”
Keith shares his “Barrels and Ammunition” framework for building effective teams:
“Most companies—once they get into hiring mode—just hire a lot of people. And you expect that as you add people your throughput and velocity of shipping things is going to increase. But it turns out it doesn’t work that way. Usually when you hire more engineers, you actually don’t get that much more done. You sometimes get less done.”
Keith argues that the reason for this is that most people in a company—even great people—are “ammunition.” But to improve velocity, you need “barrels”. He defines barrels as extremely talented people who can take ideas from inception all the way through to fully shipped product. Most companies start with one barrel (the founder). And when they add another, they can get twice as many things done per week, quarter, etc.
But true barrels are incredibly difficult to find:
“When you have them, give them lots of equity, promote them, take them to dinner every week because they’re virtually irreplaceable. They’re also very culturally specific. A barrel at one company may not be a barrel at another company.”
Video source: @ycombinator (2014)
We've released all code and models for FlashDepth! It produces depth maps from a 2k, streaming video in real-time. This was a really fun course project inspired by discussions with @mohsaied and @stevenygd and we look forward to presenting it at #ICCV2025.
GitHub: https://t.co/IHPYJEZFIj
Project page: https://t.co/dGHdXJKLKB
We've simplified local MCP usage by creating something new we call Desktop Extensions (.dxt files).
These package your local server, handle dependencies, and provide secure configuration so you can one-click share and install local servers on Claude Desktop and other apps. https://t.co/GBhHlnb8OP
This Reddit user used this AI workflow to fully automate videos on an Instagram account and got ~5M views in 3 weeks.
The ensloppification of the internet is upon us. Silicon wafers can now independently farm human attention. https://t.co/U3p6SKa133
Here are my top three pieces of advice for teams starting to build voice agents: latency and instruction following drive almost all tech decisions; you'll need fairly complex tooling to go from POC to production; start with a proven tech stack before you experiment.
I tried to write these down as simply and non-sensationally as possible. These are not hot takes! But I talk to ~10 teams a week who are getting started in voice AI, and to most people these are not obvious.
On Thursday the @covaldev team and I are hosting an in-person working session on scaling, debugging, and evaluating voice AI applications. Apply to join us, if you're building a voice application.
advice for teams starting to build voice agents