After experimenting with a "smart friend" cost control solution back in the SWE1.5 days, we put it on pause because it wasn't good enough. But this space has been on our minds and we've continued iterating.
The "sidekick" pattern in Devin Fusion has finally made the cut. Making the cut in @DevinAI today means it's pretty damn good.
Conventional model routing sucks. It passes benchmarks but fails to write code you'd actually merge.
Introducing Devin Fusion, a new hybrid-model harness for agentic coding.
In testing, it reduces the cost of Fable-level intelligence by 35% and still feels good to use.
Cognition is partnering with @MercedesBenz to accelerate software engineering across their global engineering teams, representing one of the most extensive deployments of AI software engineering in the automotive industry to date.
@ScottWu46 sat down with Katrin Lehmann, Mercedes-Benz CIO, to discuss the work:
This guide will increase your productivity by 1000%
I have been testing every harness I could get my hands on for a while, & one day I came across one that was so far ahead in terms of performance my mind was simply blown. The only con, out of the box it couldn't do everything Hermes and OpenClaw can. So I fixed that, and in my opinion created the greatest agentic harness in terms of speed, quality, & price
It started somewhere nobody was even looking. WINDSURF! Yes I know, you probably never even heard of it, or if you have, u never gave it much thought. That's where u messed up.
Windsurf is by far the fastest agentic coding experience I have ever had. So I decided to give it the capabilities of Hermes and OpenClaw as well as optimize it for even more speed & I made a guide for you so you can use it also!
I really hope this helps you guys, having a speed advantage like this for such a little price is an absolute game changer.
https://t.co/yllk4ahjKd 🫡
you can now get devin to fix issues flagged by greptile
with one click, launch a devin session that includes the issue and relevant context.
greptile and devin will loop until greptile gives the PR a 5/5
this has become the default way in which i use greptile
Let Devin work the night shift.
At 1:40 AM, @jalagar_eth told Devin to build the first NFT mint tab for @opensea mobile.
24 hours later, Devin delivered. All it took was one @linear ticket.
Introducing Devin 2.2 – the autonomous agent that can test with computer use, self-verify, and auto-fix its work. Try it for free!
We’ve also overhauled Devin from the ground up:
- 3x faster startup
- fully redesigned interface
- computer use + virtual desktop
...and hundreds more UX and functionality improvements.
We are so back.
Dropping this evening, @annanay (QFEX founder, ex-Flow/Tower quant) on why:
hard work > talent, horizontal trading ladders, and why everyone's wrong about exchange metrics
Key quote from the ep for me: "YC's $125k for 7% seemed like a lot. now we'd do it again in a heartbeat"
Full ep in bio
++ the echoing dap is my favourite part of the trailer.
Your PRs should fix themselves.
With Autofix, Devin now closes the loop on its own PRs.
If Devin Review or a GitHub bot flags bugs, Devin automatically fixes the PR. Devin also tackles CI/lint issues until all checks pass.
How to set up Autofix and Devin Review 👇
We are excited to expand our partnership with @cognition_labs to bring a full stack agentic loop to software development cycles. Cognition and Windsurf combined are a single stack for AI native autonomous software engineering with Devin’s autonomous agents sitting on top of Windsurf’s agentic Integrated Dev Environment(IDE).
Meet Devin Review: a reimagined interface for understanding complex PRs.
Code review tools today don’t actually make it easier to read code. Devin Review builds your comprehension and helps you stop slop.
Try without an account: https://t.co/Zzu1a3gfKF
More below 👇
the rate at which the Windsurf team has been shipping features + improvements + bug fixes is out of this world
Windsurf 3 months ago and Windsurf today might as well be different products
Opus 4.5 is the best model for agentic coding today, and you can now use it in Windsurf at Sonnet pricing!
This model crushes the benchmarks and also passes the “vibe test”.
Opus is also way better at instruction following, tool use when working with many tools (if you have a lot of MCP servers/tools enabled), and also isn’t too ambitious or over-eager on coding tasks.
I’ve loved using Opus 4.5 for codebase discovery/research, planning, and general agentic coding. This model paired with Windsurf’s Fast Context has been amazing for working in large codebases.
Try it out in Windsurf and let us know what you think!
Devin now has full computer use capabilities and can share screen recordings.
You can control desktop apps, build and QA mobile apps, and automate tedious work.
Here are some examples that blew our team away:
1. Making a desktop game
Love this picture from the latest @cognition blog post! A lot of what @modal is about, is really just taking dev work in the "valley of death" and shifting it left. We think about it within infra, not coding agents, but it's the same idea about keeping the flow.