fforres

And so it took us less than a month to organize it! 🔥 200+ curated folks, $60K US in total prizes, $10K cash, 36 hours of hacking focused purely on Agentic AI. hack.skyward.ai It's going to be amazing!! and it is only possible thanks to our parters & sponsors @CloudflareDev, @skyward_ai, @ElevenLabsDevs , @codeRabbit, fintoc.com and @continuumhq 💪 Winners will direct mentorship from @cen_ia. 🔥 And all the wonderful that contribute their time, $ and will to improve the tech ecosystem , @Muni_provi, @Cloudflare, @Inacapinos, @dccuchile, @DuocUC, @uchile, @CommunityOS_, @iadevsconf, indies.cl, @jsconfcl, @ElevenLabs. See you at the hackathon!

@eringriffith Glad we didn't 😮‍💨

@eringriffith for real. My co-fonder and I took a couple of meetings with them (and compai also) at the beginning of last year. Ended up not going with delve bc things seemed fishy/weird. This was.... Mid Feb 2025?

A detailed and brutal look at the tactics of buzzy AI compliance startup Delve "Delve built a machine designed to make clients complicit without their knowledge, to manufacture plausible deniability while producing exactly the opposite." substack.com/home/post/p-19…

@JayaGup10 @akoratana Love this :)

DM me or @akoratana

@mattzcarey Agreed. Though a limit of 6 concurrent outgoing connections has been tough. Can that be higher?

Dynamic Worker Loaders with GlobalOutbound Worker from a Worker deployed by a Worker for Platform

What in the automated AI driven spam hell is this shit. FYI @github

"See this? This is where Claude fucked up"

@peterschacherl @PlanetScale Right? So friggin fast

Moved an app to Cloudflare from self-hosting because of performance and the need for high availability. And since I'm using Cloudflare stuff like R2, I can simplify my life with bindings. Also, @PlanetScale via Hyperdrive is insane.

@levan @Cloudflare @PlanetScale @zero__ms on the cloudflare dashboard, it's the hyperdrive analytics section

@fforres @Cloudflare @PlanetScale @zero__ms where do you track this?

Average latency in @Cloudflare workers from our previous AWS DB, to @PlanetScale From 255ms, to 10ms (connection latency is also reduced, from 151 to 3.7ms) Combined with @zero__ms, it's chef's kiss

@zeeg We were going to cure cancer & solve fusion.

Remember when labs said they were focused on AGI? Remember when they were research focused and non-profit? Weird how they all just look like enterprise software companies now 🤑

@_ashleypeacock A man can only dream 🙏

@fforres Not that I’ve seen!

Here's a recap of EVERYTHING shipped by Cloudflare in February, it can be hard to keep track as there's always so much new stuff dropping: 🤖 Agents SDK - CodeMode: A complete rewrite into a modular, runtime-agnostic SDK to improve how AI models orchestrate tools - Workflows integration - scheduleEvery() method for fixed-interval recurring tasks - Readonly connections for spectator views and custom MCP OAuth providers for advanced authentication - built-in retry utilities with exponential backoff and per-connection protocol message control for WebSocket clients - RPC transport for MCP, allowing Agents to connect to McpAgents in the same Worker with zero network overhead 📦 Containers - Significantly increased concurrent instance limits to 6TiB of memory, 1,500 vCPUs, and 30TB of disk (15x increase!) - Docker-in-Docker support added 🚀 Workers - Workers are no longer limited to 1000 subrequests per invocation - Windows Support for Python Workers - Better support for React Server Components (RSC) during local development - Best practises guide now available 🎁 Sandboxes - Create and restore snapshots of sandbox environments to skip setup steps - Interactive browser terminals now available 🧑‍🔧 Pipelines - Dropped event metrics to identify schema mismatches - Typed TypeScript bindings, - Simple setup mode for R2 buckets 🧠 Workers AI - TanStack AI adapters for Workers AI and AI Gateway - GLM-4.7-Flash model on Workers AI 🗄️ R2 - Local Uploads in beta, improving global upload performance - R2 SQL now supports approximate aggregation functions 🔐 Workers VPC - Support for Cloudflare Origin CA certificates 🔁 Workflows - See visual logic diagrams for your Workflows directly in the dashboard 📨 Queues - Now available to all accounts on the Workers Free plan, including 10,000 daily operations. 🔍 Observability - Write structured queries to filter and search your Workers logs and traces ⚡️ Wrangler - Deploy now auto-detects your framework, no need to create a Wrangler configuration file 🔎 AI Search - Improved performance for more granular indexing ⏰ Durable Objects - deleteAll() now clears alarms (requires compatibility flag!) I probably missed something, but did my best to get it all into one post and without misrepresenting anything!

@programmerByDay @Cloudflare @PlanetScale @zero__ms probably fast :) (though we do get in diminishing returns area at this time, 5ms more/less is not that big of a deal at this time) D1 is great main thing is that we need postgres as @zero__ms is core to our product

@fforres @Cloudflare @PlanetScale @zero__ms Interesting. I’m wondering how muc would that be if you use cloudflare db

@zebassembly @craigsdennis @Cloudflare @PlanetScale @zero__ms We have, it has really helped. :)

@fforres @craigsdennis @Cloudflare @PlanetScale @zero__ms Were you using the new region support? You can make your Worker run next to an AWS/GCP region

@m0rcutt @Cloudflare @PlanetScale @zero__ms Its so good 🤤

@fforres @Cloudflare @PlanetScale @zero__ms Essentially did the same thing. Still have some baggage on AWS but have the rest on @Cloudflare, @PlanetScale, and @zero__ms. It is quite great

@TheBuilderJR @Cloudflare @PlanetScale @zero__ms Same region, same workers, same codebase. top of my head, aws requires a few things to wire up everything safely, no doubt you people with time can make it fast AF. (I mean... Planetscale itself is proof).... But time is the key word

@fforres @Cloudflare @PlanetScale @zero__ms wow do you have intuition on what made the difference? is the regional distance similar?

@jonas @PonziChad @harshil1712 @Cloudflare @PlanetScale @zero__ms No caching

@fforres @PonziChad @harshil1712 @Cloudflare @PlanetScale @zero__ms Are you caching queries in hyperdrive at all?

No, we are hosting it on their own cloud. We experimented with a full cloudflare hosting on containers some months ago, but between forking litestream to work with r2, and the size of containers available at the time (and uptime of instances) it was not good. Granted, this was when containers were in alpha (like before they were released), container specs are a lot better now :) (maybe litestream has support for R2)

@fforres @aboodman @Cloudflare @PlanetScale @zero__ms Are you running the change streamer in a worker? 👀👀👀

@_zen_garden_ @craigsdennis @Cloudflare @PlanetScale @zero__ms Yep, we have product requirements that sadly could not be achieved with sqlite (not without very sizeable engineering investments on our end)

@fforres @craigsdennis @Cloudflare @PlanetScale @zero__ms Sounds like you were not using D1 sharded but you had a central postgres db????

@PonziChad @harshil1712 @Cloudflare @PlanetScale @zero__ms Both using hyperdrive

@fforres @harshil1712 @Cloudflare @PlanetScale @zero__ms With hyperdrive?