Teams at Canva, Uber, and Databricks are building AI agent workforces with @RelevanceAI_. Its workspace puts docs, apps, and automations on the same filesystem, giving agents richer context with less token overhead.

Cloudflare tried off-the-shelf AI code reviewers. We needed more control, so we built our own. It's now run over a million reviews. @_ryanskidmore helped make the build-vs-buy call and shares how teams should weigh control, cost, and maintenance in their decisions.

Bitcoin’s halving happens on schedule. Price doesn’t. Past cycles lined up closely with the 4-year halving rhythm, but the current cycle has already broken old patterns. So is Bitcoin’s 4-year cycle still intact? Here’s the breakdown 👇

The next prompt injection your coding agent reads might be an error report. So Sentry screens for it first, spending $263 a day on Jev to check 10M events. Now the team is testing its own classifier, aiming to match Jev's accuracy for 90% less. @grichadev explains how.

Shared infra works great until agents hammer your app. Then, one overloaded DB can take everything down. That's where Durable Objects come in. Split the load across as many tiny, SQLite-backed servers as you need. Learn how, plus 4 mistakes to avoid, from @AshleyPeacock.

What do you do when a customer needs 1,000 Git repos created per second, and you have three weeks to deliver? You build your own Git server on Durable Objects. @dillon_mulroy breaks down how a small but mighty team of three shipped Cloudflare Artifacts as a closed beta, on time.

When you're authorizing card payments, every millisecond counts. That's why @tryramp moved its routing to the edge. Their Rust Worker reached 100% of traffic within weeks and now handles 500M requests and 10 TB daily. Head of Production Engineering @raphaelbeamonte shows how.

Agents that run for hours can't afford to lose their place. Standard Agents gives every conversation its own Durable Object and database. At Cloudflare Connect, co-founder and CTO @jpschroeder shows how it works: - One database per conversation, because agents reread their full history on every step - Alarms kick off long work, and a watchdog object tracks progress so agents keep running for hours - A live demo where every attendee who scanned in got their own expense agent