
@cognition 𦦠/ upenn m&t / prev. ai @figma, @kp_fellows
Joined July 2017
- Tweets277
- Following476
- Followers1.2K
- Likes2.1K
kylie chang π’πΈ retweeted
Our Brand & Creative team at @cognition is growing. Iβm looking for:
β Video Producer, Customer Stories
β Art Director, Campaign
β Copywriter, Campaign
β Video Editor, Contract
β Brand Designer
β Brand Designer, Events
β Motion Designer
β Production Designer
This is a really special moment here, and I'm looking for people who are excited to build the creative studio they've always wanted to be part of. I canβt respond to all DMs, so please reach out to our amazing recruiter Monica Jin on Linkedin (link in thread).
devindash & devindance πΊπΊπΊ
β¨ Introducing DevinDash β¨
Starting today, dash food with the DoorDash MCP and a Devin Bot.
Startups can claim $1k of free DoorDash: devin.ai/startups
kylie chang π’πΈ retweeted
There's a lot of data in tennis.
Here's how seven-time Grand Slam singles champion and four-time Olympic gold medalist @Venuseswilliams uses Devin Bots to make sense of it all.
kylie chang π’πΈ retweeted
Bots are super cool and if you haven't used them yet, you absolutely must try them. Create a bot to manage a goal or a project that you'll work on across multiple sessions and days.
Some technical tidbits about the architecture and design:
Devin used to have a major UX problem: a session today doesn't remember exactly what happened in a session from a few days ago. It only has vague awareness of those previous sessions, and recalls only the nuggets of information that were explicitly persisted. This is because each new Devin was essentially an ephemeral clone of a base template. One symptom of this problem is that our users would reuse the same session instead of starting new ones, since they didn't want to lose context, and end up with tens or hundreds of PRs from a single session.
Bots almost completely solves this problem of context being lost between sessions, because now inter-session context is surfaced by default. Teams now create dedicated bots for different projects; while these different devins can talk to each other when necessary, they each have their own individual memory bank and goals. Each bot now has its own, long lived identity.
The transition from "Devin is a swarm of ephemeral agents" to "Devin is a collection of bots, each with their own persistent identity and memory" is changing how people use and prompt Devin. I see people investing much more in providing high level project context to their bot, and requiring fewer messages for each additional task since it already has a ton of background context on the project.
This is the most significant change to how Devin works since our original launch 2 years ago. And it in turn changes how you use Devin too: instead of delegating tasks, you should delegate goals and responsibilities to Devin.
It's crazy to think that all this happened because we changed how memories get persisted and retrieved. With each bot having its own memory bank, and a given org having hundreds or thousands of bots, we can now be much more aggressive in surfacing memories without worrying about polluting the agent's context with irrelevant memories. The even crazier part is that this is almost entirely an orchestration change, and that while model improvements played a role, we probably could've implemented this a year ago and observed similar effects.
Anyway -- we've come a long way from the retrieval and reranking strategies of 2024, and the result is Devin Bots -- probably the most ergonomic and futuristic UX for coding agents I've ever used.
devin bots can take someone from 50 PRs a week to nearly 2,000. they can also own long-horizon outcomes, like keeping me hydrated and checking on my posture every afternoon. so excited this launch is finally out! π¦
kylie chang π’πΈ retweeted
Introducing: Devin Apps
Private, secure, sharable web applications running directly on Devinβs cloud computer.
Devin Apps are designed to power your most ambitious internal tools, preview deployments, and extensions to Devin itself.
kylie chang π’πΈ retweeted
Introducing SWE-2 Flash, our new fast & efficient coding model.
On leading evals, it outperforms all American open-weight models, even ones 3x larger.
To balance transparency and cyber risks, we are building a program to securely share the weights with partners & researchers.
kylie chang π’πΈ retweeted
Founded by three 22-year-olds in 2023, @mercor crossed $2B ARR in June, only 4 months after hitting $1B.
Today, millions of experts on the platform bring their knowledge of medicine, law, engineering, and beyond to frontier AI - helping models take on increasingly complex and consequential work in the real world.
Mercor's engineers move at extraordinary speed.
They give Devin long-running tasks before bed, during meetings, or on their way to the gym. They come back to tested features and PRs ready to merge.
kylie chang π’πΈ retweeted
Introducing FrontierCode 2.0: our new coding benchmark that measures a modelβs ability to go beyond the spec.
Most coding evals measure a modelβs ability to write code that meets a spec. With FrontierCode 1.0, we asked a new question: would you actually merge the PR?
FrontierCode 2.0 is comprised of more diverse tasks, featuring larger scopes and higher mergeability standards. We are now asking: not only would you merge this PR, but should you?
One key feature of FrontierCode 2.0 is underspecified prompts. Rather than including every engineering decision, prompts are deliberately ambiguous, giving the models room to demonstrate their judgment.
The result is a benchmark that is far from saturated, with the top model scoring just 45%.
kylie chang π’πΈ retweeted
We are entering a new dark age of cybersecurity. Devin is here to help.
Introducing the new Devin Security: powered by Claude Mythos and frontier cyber models, plus hundreds of millions of dollars of credits to help teams find, prioritize, and fix urgent vulnerabilities.
kylie chang π’πΈ retweeted
Weβve open sourced an agent plugin that helps secure systems against recent cybersecurity threats.
You can install them with any agent or use them in Devin today:
devin.ai/blog/vulnerability-β¦
kylie chang π’πΈ retweeted
Today we're introducing a refreshed Devin Desktop app, with an agent command center designed to be smoother and more polished than ever before.
kylie chang π’πΈ retweeted
18 million miles traveled, 400 red hoodies, one 260-digit number, and 130+ former founders.
We asked some of the people building Devin to tell us how they got here and what life at Cognition is really like. Hereβs what they shared.
cortisol levels have never been lower since devin oncallπ§ββοΈ
kylie chang π’πΈ retweeted
Introducing Devin in Pylon: Accelerate your Median Time to Fix (MTTF), with $15,000 in credits for new Devin customers.
kylie chang π’πΈ retweeted
We're also open-sourcing our memory format:
- support graph relationships
- updates over time with historical records
- backed by git and markdown
and you can use it with any agent, not just Devin!
Introducing Dreaming: Across sessions, Devin builds a memory graph of how you like to work
At night Devin self-improves its memory to remove stale records and discover latent information
We are creating an OSS standard called Agent Memory Repo: cognition.com/agent-memory-rβ¦


















