Pinned
building @cognition
San Francisco, CA
Joined February 2023
- Have had a blast building Devin - very grateful to @ScottWu46 and the entire team for all the support! (for those wondering, yes I did end up graduating and my parents are very happy with my decision to join Cognition ๐)When you hear the advice "hire the best," this is what that looks like: "We had a candidate who came and interviewed. He was a junior at MIT and very very young. We gave him our interview and he did way better than almost any of the full-time candidates that we had ever talked
00:00 - Devin is now faster, cheaper, and smarter than ever before. With Devin Search, interactive planning, cloud IDE and much more, Devin 2.0 closes the loop on harder tasks from brainstorming -> planning / specification -> execution -> testing + review. Available now starting at $20!Introducing Devin 2.0: a new agent-native IDE experience. Generally available today starting at $20. ๐งต๐
00:00 - Cognition ๐ค Windsurf. Best team up ever.Cognition has signed a definitive agreement to acquire Windsurf. The acquisition includes Windsurfโs IP, product, trademark and brand, and strong business. Above all, it includes Windsurfโs world-class people, whom weโre privileged to welcome to our team. We are also honoring
00:00 - Replying to @que_touristCan't wait for the tourist vs Devin exhibition match
- My new daily flow is to get Devin to scope all my tickets and then have Devin implement the ones where it had high confidence. Another big step towards having Devin integrate into all parts of the engineering workflow!We built a @linear integration! Now you can run Devin on your entire ticket backlog. Our Devin x Linear workflow ๐
00:00 - Magical work by @silasalberti (and many others) to make Devin snapshotting + wake up times nearly instant!We needed instant VM snapshots for Devin but EC2 took 30+ minutes. So, @silasalberti built blockdiffโa new file format that makes snapshots 200x faster. Today, weโre open-sourcing blockdiff & sharing how it works ๐๐
- Great write-up by @walden_yan on how we think about LLM context at Cognition. Must-read for anyone building agents.Weโre sharing more about how we engineer agents. @walden_yan on why you shouldn't build multi-agents ๐๐
- How DeepWiki works under the hood, in 2 minutes ๐น For more details on how we build Devin, check out @russelljkaplan's full talk at @langchain Interrupt ๐๐
00:00 - One of the underrated benefits of Devin imo is that it doesn't just save you time, it saves you mental energy. Often times working with Devin to root cause a bug or map out a feature is just as valuable (if not more) than having Devin actually implement the changes.The new devin experience is actually sick. Saved me a lot of mental burden the past few weeks!
- One of my favorite workflows with Devin is multiplexing with 5+ Devins while eating or traveling, boosting me from 0 -> 5x productivity. Easier than ever now with Devin's cloud IDE + enhanced interactivity.Replying to @cognition and @cognition_labsthe future of agents is parallelization!
- Devin ๐ค Roxagent + tools? so yesterday. weโre in our group chat era now โ agents talk to agents ๐. w/our bffs (and beloved customers) โ @cognition_labs โ weโve developed an agent-to-agent protocol that lets the rox agent swarm hit up @DevinAI to turn sales calls into instant demos. itโs
00:00









