Capability, reliability, and UX
Perceived improvements in agent quality, speed, autonomy, tenacity, interaction design, friction, and remaining limitations.
43.3%
Best tweets about Devin
30 Best Tweets About Devin (2026)
Discover the best tweets about Devin AI, including software engineering tasks, coding workflows, benchmarks, product updates, team adoption, and limitations.
Cognition Devin software engineering agent capabilities, coding tasks, benchmarks, releases, adoption, reliability, limitations, and user evidence.
- Creators
- 24
- Updated
Original Xholic analysis
What 30 top Devin posts reveal
The conversation presents Devin as a cloud-oriented engineering agent used for PR workflows, browser testing, review, and multi-agent orchestration. User accounts are positive overall, but individual posts also identify environment setup, review throughput, token consumption, usage limits, and confidence in task completion as practical considerations.
- Dominant tone
- Positive
- Median score
- 12.3
- Leading format
- Case Study
- Recent posts
- 53.3%
70% of posts
All-time engagement
36.7% of posts
Published in 90 days
Conversation map
The themes creators return to
Adoption, productivity, and ROI
User reports and company metrics on shipped work, merged PR volume, engineering output, cloud-agent adoption, pricing, usage limits, and return on token spend.
36.7%
Testing, code review, and security
Browser and end-to-end testing, automated PR review, edge-case detection, supply-chain defense, bug fixing, and reviewer trust.
30%
Autonomous delivery and PR workflows
Agents implementing features, testing, opening or merging pull requests, responding to GitHub discussions, and producing review artifacts.
26.7%
Cloud-native agent workflows
Using Devin as a long-running cloud engineer integrated with repositories, Slack, Linear, CI, browser testing, and remote environments.
23.3%
Benchmarks, models, and inference
APEX-SWE evaluation, SWE model releases, model behavior and speed, token limits, and comparisons among coding-agent stacks.
13.3%
Execution environments and integrations
macOS/iOS support, computer use, Outposts on private infrastructure, sandbox setup, MCP integrations, and Devin Desktop.
13.3%
Multi-agent orchestration
Devin dispatching and managing subagents, middle-manager patterns, task decomposition, parallel execution, and context management.
13.3%
Tone and stance
Sentiment Positive leads
Author posture Supportive leads
Performance benchmark
- Median likes
- 77
- Median reposts
- 6
- Median replies
- 7
- Median views
- 6.7K
Posts with media make up 56.7% of this collection. Their median all-time score is 23.3, compared with 2.97 for text-only posts.
Format mix
- Case Study 36.7% · score 12.7
- Announcement 33.3% · score 26.1
- Opinion 16.7% · score 4.64
- Prediction 10% · score 12.0
Consensus and debate
Where creators agree, and where they do not
Shared view
Cloud workflows support long-running engineering processes
Posts describe cloud-based workflows that connect agents with tools such as Linear, Slack, browser testing, and CI. One user describes automated merging after tests and CI pass; another describes an orchestrated workflow that produced more than 60 ready-to-merge PRs over 12+ hours.
Shared view
Review and testing are recurring use cases
Posts report Devin identifying edge cases in review, catching an axios supply-chain attack for Cognition customers, and generating a screen recording and test report in a macOS/iOS demo. These are user, company, and product-demo reports rather than independent evaluations.
Shared view
Delegation and review feature prominently in accounts of agent-assisted development
Several posts describe a workflow in which people specify work, agents implement or coordinate it, and humans review results. One post explicitly frames its workflow as delegation rather than autocomplete, while another describes Devin agents starting other Devin sessions.
Open debate
Autonomy versus confidence in task completion
One user describes Devin as unusually self-directed after completing two design-engineering tasks overnight, while another says the SWE-1.7 harness is not sufficiently tenacious to inspire confidence that it will finish a task.
Open debate
Cloud-workflow value versus operational friction
One user calls cloud-based work through Slack and Linear a “massive unlock” and reports increased shipped work, while another says cloud-agent adoption has been slower than expected and that setting up and managing cloud environments remains difficult.
Open debate
Automation throughput versus token and usage efficiency
One post describes a 15-line feature taking two hours and an estimated 1 billion tokens across testing and review loops, while another reports hitting GPT-5.5 limits in Devin Desktop and getting more work done with Opus in Claude Code.
What performs
Patterns behind standout posts
Media posts had a higher median score than text-only posts
Deterministic analytics identify 17 media posts (56.7% of the 30-post set). Their median all-time score was 23.292, compared with 2.97 for text-only posts. Two high-scoring media-associated examples covered macOS cloud agents and Devin Review’s reported supply-chain detection.
The highest-scoring post documented a cloud orchestration playbook
The highest-scoring post, at 346.74, described a cloud “software factory” workflow. Its author reported 60+ ready-to-merge PRs, 12+ hours of operation without blocking, and review capacity becoming the bottleneck.
Security-review and macOS-support announcements were among the largest outliers
The post reporting Devin Review’s detection of the axios supply-chain attack scored 211.77, and the macOS cloud-agent announcement scored 216.64. Both were among the five all-time-score outliers identified in the deterministic analytics.
Statistical standouts
- View standout post 1 Score 346.7 · 28.12× median
- View standout post 2 Score 216.6 · 17.57× median
- View standout post 3 Score 211.8 · 17.18× median
- View standout post 4 Score 164.7 · 13.36× median
- View standout post 5 Score 145.0 · 11.76× median
Creator landscape
Who shapes this conversation
The five most represented creators account for 33.3% of the selected posts.
-
1. nader dabit
@dabit3
2 posts
-
2. Nicholas Charriere
@nichochar
2 posts
-
3. Ryan Carson
@ryancarson
2 posts
-
4. Scott Wu
@ScottWu46
2 posts
-
5. Vincent van der Meulen
@vinvan
2 posts
-
6. Walden
@walden_yan
2 posts
Repeat contribution was concentrated among a small set of accounts
The dataset contains 24 creators, and deterministic analytics attribute 33.3% of top-five placements to the top five. Vincent van der Meulen, Ryan Carson, Scott Wu, and Nader Dabit each contributed two included posts covering workflows, adoption, review, or releases.
Specific workflow accounts supplied detailed usage evidence
Posts with concrete operational detail describe browser testing and PR merging, GitHub-thread behavior, and cloud-based work through Slack and Linear. These accounts are more specific than general product-positioning posts, though they remain individual reports.
Since the previous snapshot
What changed since Aug 12, 2026
- 96.7% of the selected posts remained.
- The creator count changed by -4.
- The leading sentiment remained stable.
How this analysis was made
Themes, sentiment, stance, and post format are classified per tweet. All counts, shares, medians, creator concentration, freshness, and performance comparisons are then calculated directly from the published snapshot.
Xholic's all-time score compares engagement while accounting for reach, post age, and creator consistency. It is used for relative comparisons within this collection.
This report analyzes the exact 30-post snapshot shown below. AI identifies editorial categories and drafts explanations; all statistics are calculated from the snapshot, and every narrative claim is checked against cited posts before publication.
Best Devin tweets
Top Devin tweets from 24 creators
Ranked 01–30
- 01
@vinvan ·
fable produced 60+ (!) ready to merge prs for @mainframe overnight. here's how you can set up a software factory using fable as an orchestrator. prompt included! 1. narrate your *entire* todo list — everything that's top of mind — using dictation. i used @usemonologue and my prompt was ~10K words! you have to use voice because we want excruciating levels of detail. 2. locally, tell fable to plan all the work and put it in @linear. the goal is to make it such that a dumb model dropped in linear knows *exactly* what to do. aka all issues should have a clear DAG, plan, and optimize for max. parallelizability. i recommend running fable on high and having it fan out to codex subagents on xhigh (using the cc codex plugin.) to save usage. just make sure that fable itself is doing the architecture for the hardest issues! 3. now switch to @DevinAI. we'll use devin to run our factory because our orchestrator needs to be a cloud agent (it's going to run for a long time!), and be able to spin up other cloud subagents. 4. hook up the @linear @SlackHQ mcps. this will let your agents read the fable planned work, and communicate with you/your teammates. e.g. it can talk to your coworkers to make sure it's not stepping on anyone's feet! 5. a good feedback loop is *crucial.* before starting, make sure you have a bug bot (we use @cursor_ai.) and that your agents can test all its changes. e.g. we use the fantastic @limrun to make sure our agents can do ios development e2e. 6. create a devin ultra (fable) agent — this will be your orchestrator or 'middle manager.' we're going to tell it to look at linear and dispatch subagents for all tasks. the middle manager's goal is to keep spinning up subagents with the right level of intelligence until all the work is done. and keep all agents unblocked while you're sleeping, hanging out in the park, whatever! the reason the middle manager is able to orchestrate all of this is because fable planned everything in linear. and because the middle manager itself is also fable (contrary to most of its subagents.) 7. to prompt your middle manager, attach the middle manager prompt i'll post in the comments, modified for your repo. it's important that we *attach* the middle manager's prompt to our conversation vs. typing it out, so it never gets compacted. the middle manager's context window is holy. so our prompt tells it to spin up subagents — instead of doing things itself — for pretty much everything. 8. the prompt has some other cool, non-obvious details. firstly, every agent is told to keep working until it's met the burden of proof. in our case that's a (self)-review, bug bot being green, bug bot fixes being elegant, and a recording. we tell our middle manager to enforce this. secondly, we ask the middle manager to make sure linear stays up to date and communicate via the slack mcp. specifically, it should tag coworkers in slack whenever it has a question and post heartbeat updates in the eng channel. 9. your software factory is ready to go! in our case, it worked for 12+ hours without getting blocked. and produced extremely high quality prs. addendum: all of this is possible because with @AnthropicAI fable, models are finally smart enough to orchestrate/navigate large sets of work and their dependencies. and fable is a great software architect! now of course the 'downside' is that you're now review bottlenecked. we have 60+ PRs to review! but i think as an industry we should be able to come up with some innovation there. we'll certainly try our best to help here with @mainframe too! hope this is helpful! happy to answer any qs.
- 02
@dabit3 ·
You can't build iOS apps without a Mac, so we gave Devin one. Devin Cloud Agents now run macOS, with Xcode, iOS simulator, and your signing setup, all in a real macOS environment (with full computer use). In this demo, Devin builds a native iOS game, then plays and tests it with computer use. Devin then sends a screen recording of the gameplay along with a test report for my review. With @namespacelabs you get as many macOS environments as you'd like, configurable with many major versions of macOS operating system and Xcode.
Watch video
- 03
@ScottWu46 ·
Devin Review caught the axios supply chain attack for multiple Cognition customers before the attack was publicly known. These attacks will be 10x more frequent in the age of AI; it is critical that repo maintainers start using AI for defense as well. (showing one example below where Devin Review caught the attack within an hour of its release - text minorly edited for anonymization)
- 04
@adarsh_exe ·
Traditional coding benchmarks do not reflect how software is actually built and maintained. That's why we built a new benchmark, APEX-SWE, in partnership with @cognition. It measures whether AI models can perform complex, real-world software engineering work to ship systems that work and debug them when they don't. @OpenAI GPT 5.3 Codex (High) tops the leaderboard at 41.5% on Pass@1.
Watch video
- 05
@ryancarson ·
This is how I’m currently running my startup with @DevinAI + @openclaw The browser testing in Devin is mind-blowing. I was trying to duct tape and jerry-rig all this stuff together with Playwright + uploading videos to PRs and all sorts of stuff and Devin just does it all e2e. Wild. I didn't show how I'm using @linear, which I am using for issue tracking. I have a "land" skill that I tell Devin to use whenever all the browser testing is done and all the CI goes green and it just merges the PR
Watch video
- 06
@vinvan ·
some reflections from solely using cloud agents this year: 1. every engineer should default to cloud. it completely changes how you view and use agents. if you run a company, it might be worth mandating everyone starts in cloud 2. cloud agent adoption has been much slower than i expected— e.g. looking at a ton of cursor profiles it’s clear majority cloud usage is still rare 3. getting your dx cloud agent ready still requires creative jiu jitsu. dev infra docs could be much better — “this is how to make our stuff accessible to agents/parallelizable.” luckily investments also benefit humans 4. it’s still a PITA to setup & manage cloud envs across cursor/devin etc. but i assume it’ll get bitter lessoned and we don’t need conventions for setup scripts etc. 5. where are the labs?! would love to see codex et al. invest more in their cloud experience. i know they can do it :) 6. it’s strange that cursor/devin’s investment in mobile apps lags behind their investment in cloud agents. they should go hand in hand. the ability to start agents from slack mobile isn’t enough! 7. a cloud agent spinning up other cloud agents (middle manager pattern) is goated. e.g. nice to go for a run, yap for twenty minutes, and end up with parallel agents. only devin supports this well 8. the uis of ADEs have somewhat adapted for cloud agents. but ui patterns for upcoming long running *and* proactive agents are understudied. super excited to see more experiments here (and will contribute) overall: i freaking love cloud agents. you’ll dissappoint me personally if next month you still spin up more local agents than cloud. very grateful for cursor and devin for making this technology so easy to use!
- 07
@cognition ·
Introducing Devin Outposts: run Devin on any machine. Your Mac mini, a GPU box in your lab, a VM inside your private network, or a Kubernetes cluster next to your internal services.
Watch video
- 08
@ryancarson ·
I haven't typed `npm run dev` on my local machine for three days now and it's absolute bliss. Having my agents 100% in the cloud is a massive unlock. (One of those agents is openclaw, which is technically on my mbp in my office, but the only way I interact with it is via email/slack so it “feels” cloud) I'm able to run all the engineering and marketing for my startup through Slack and Linear and because of this the work product that I'm shipping has increased dramatically. I know all of us devs love creating our own custom solutions to this stuff but the truth is that creating an agent orchestration layer for your company or startup is a full-time job. Our job as startup founders is to be growing the company, not to be building out an agent orchestration custom platform. I think if you have a larger engineering team like Ramp, then it does make sense to build an entire layer like Inspect agent. However, I would venture to say that I'm getting most of the value by simply paying for a pre-built, battle-hardened solution like Devin. Again to be clear I'm not being paid by Devin or anybody to say these things, just my real-world experience using this stuff.
- 09
@AmOptimistShow ·
EPISODE 147: The New Era of Software Abundance @JTLonsdale visits @ScottWu46 & @russelljkaplan at @cognition HQ 00:00 Episode intro 01:35 Why technical talent & execution matters in AI 06:10 Do young people have an edge in the AI era? 08:26 Cognition's rapid growth 11:55 The new era of software abundance 14:30 Cognition engineers don't type code anymore 19:20 "Never sleep while Devin is idling" 21:25 The case for AI disinflation 23:50 How Devin generates 12X productivity gains 28:25 Cognition for government / taking on complex, broken systems 36:40 The AI race / competition with Anthropic 39:00 Forward deployed engineers? 43:40 How fast are LLMs improving? 47:10 The AI-led small business explosion
Watch video
- 10
@dabit3 ·
New video — Devin in 8 Minutes Everything you need to know to integrate and build with @DevinAI remote agents. From creating and automating sessions to autofixing bugs, autoreviewing PRs, and connecting with Slack. Also covers: DeepWiki, Playbooks, and the MCP Marketplace.
Watch video
- 11
- 12
- 13
- 14
@MollySOShea ·
BREAKING: Devin now writes ~95% of Devin's code Scott Wu (@ScottWu46), CEO of @cognition says the abundance era of AI is real & coming soon.. FULL INTERVIEW Cog Stats: › $2.5B raised, $26B valuation › $500M+ run rate, up from $37M LY › Customer usage is up 11x-12x in 6 months › Total code shipped inside Cognition has grown 7x › Devin writes ~95% of Cog's code, up from 89% at May Series D › The entire Windsurf acquisition was negotiated over 1 weekend, ~40 people buying a 200 person company with $82M of ARR › Goldman Sachs, Mercedes-Benz, Citi, Dell, Santander, NASA, the US Navy & the US Army all run Devin R I P : "The tokenmaxx era lasted from January to May of 2026, roughly." "I'm excited to have AI that allows me to spend all of my time & focus on the things that I care about & take care of the rest. I think the abundance era is real, & I think we'll be there pretty soon." Recorded 7 July 2026. On 23 July 2026 Cognition acquired Poke maker The Interaction Company 𝐓𝐈𝐌𝐄𝐒𝐓𝐀𝐌𝐏𝐒 (00:00) Scott Wu, Co-Founder & CEO at Cognition (01:00) The Identity crisis of building Devin (01:42) How Cognition 7X-ed its output in 6 months (02:55) The Metric every AI company gets wrong (04:18) Cognition's 12x growth curve (07:46) The Windsurf acquisition (13:26) Scaling overnight: Cognition's hardest week (15:11) The strategy behind the integration of Cognition products (17:23) Cognition's M&A playbook (20:01) Why the DMV is Cognition's next big target (22:59) Why Cognition bet against AI centralization (25:09) Why enterprises refuse to bet on a single AI lab (28:32) Scott's honest take on Superintelligence (32:36) Is Silicon Valley losing its mind over the AI gold rush? (33:26) Devin's origin story (37:17) The question Scott thinks nobody's asking yet (39:59) Poker with Peter Thiel & Napoleon ? (41:24) Scott's vision for the next year
Watch video
- 15
- 16
@petergyang ·
"If you want to build great agents, you have to let the model cook." Here’s my new episode with @imjaredz (Builder in Residence at @cognition), where he shared what he learned from watching the best engineers build with AI agents, including: → Why fewer rules make agents work better → How to get agents to check their own work → Live demo: Build one agent to manage an agent team Some quotes from Jared: “Our goal is not for you to token max. Our goal is for you to ROI max.” “The trick is to break up the work to keep the context window small for each sub-agent.” “More Devin sessions now start from other Devins than directly from humans.” 📌 Watch now: https://t.co/jP1GoKtMnX Thanks to our sponsors: @RiversidedotFM: All-in-one AI studio for podcasts and video https://t.co/uWnS6aiMPE @WisprFlow: 4x faster than typing with your voice https://t.co/oqHJ8bN3ll
Watch video
- 17
@andrewdfeldman ·
Developers deserve AI tools that don't make them wait. Cognition delivered just that. Congrats to @ScottWu46 and the entire @cognition team on the release of SWE-1.6. They both reduced bad model behaviors and dramatically increased good behaviors. Developers now stay in flow. With no waiting and no friction. Cognition were one of the first to see how important fast and frictionless inference would be for coding agents. We're proud to power SWE-1.6 together at 950 tokens per second on @cerebras. Programming at the speed of thought.
- 18
- 19
- 20
@peer_rich ·
at the start of this year i made a prediction: by the end of 2026 i wont be writing code by hand im already at 80% the remaining 20% feels more like habit than necessity the way i use AI for coding now is less like autocomplete and more like delegation i describe what i want, claude code or devin handles the implementation, i review that fundamentally changes what the job actually is the engineers who will be most valuable in 2 years will be the ones who think clearly about systems, catch bad output fast, and know when the AI is confidently wrong
- 21
@markfenner ·
Someone asked why I use both Hermes and Devin. Fair question. They don’t do the same job. Hermes is my daily driver. It runs the cron jobs, business ops, content workflows, study material, and all the work that needs to keep moving whether I’m at my desk or not. It’s the operating layer. @DevinAI is where the building happens. Repos, features, PRs, security scans. When something needs to be engineered, scoped, and shipped, that’s Devin’s lane. Hermes runs the operation. Devin handles the engineering. Beyond that, I’m just a tech kinda guy. I’ll use Codex, run local models, and test whatever interesting tool dropped that week. Could I force one tool to do everything? For sure . But, I don’t need to. I’ve tuned this setup around how I work, I enjoy using it, and it’s the most efficient version of my day I’ve found. I’m always curious how other people split this work. What does your setup look like?
- 22
- 23
@ChaiWithJai ·
When I use Claude Code and Codex, its like I'm working for them because of so much steering needed. Devin feels like the first coding agent that works for me. Yesterday, I set up the sandbox environment and went to the beach and it knocked out 2 important design engineering challenges: 1. Bringing a semi-complex AI harness to life I teach critical thinking and AI skills to first-time entrepreneurs. And I want my learners to use a whiteboard to scaffold their thinking BEFORE and AFTER each AI session. My goal when teaching cognitive strategies is to allow my learners to walk away with a model of reality (i.e. mental model) that they continually test and update until it predictively explains and reconciles reality. 2. Re-design the homepage and get the core JTBD done working to onboard my first 10 users. Devin worked from 2 PM until 3 AM last night and I'm going to plan a longer retro to understand what I can rely on Devin for and what work I need to have ready for it for the application to go live. Today, I'm going to continue working on this first milestone of getting my first 10 users. My milestones are: a. Get 10 users b. Ensure the JTBD is complete for my AI World Cup Challenge c. Extend the web application into a mobile application and publish before NYC Tech Week (bonus): raise a round of funding so I can build Khan Academy for Adults I'm so happy to welcome Devin onboard to our team so I can fire myself as our engineer and get back to doing what I love, teaching. Teaching is the love of my life. @dabit3 and @cognition thank you, thank you, thank you so much. I really feel like my dreams are in reach. Its been less than 24 hours. And I want to wait to experiment more and distill why this feels so different. Stay tuned! Life is so rich, Jai P.S. I'm excited that I might have free time again! In which case, I'll be vlogging my experience trying to bring my dreams to life. What's your dogged pursuit?
- 24
@theinformation ·
Cognition is overhauling Windsurf into Devin Desktop, a hub where developers can manage AI coding agents from OpenAI, Anthropic and others. The strategy positions Cognition as a neutral platform in a market increasingly dominated by model providers. Full story: https://t.co/ZmPZ4t1PKJ
- 25
- 26
@boyuan_chen ·
Return on tokens is becoming a real product surface. Cognition says its typical customer saw agent usage grow 1000% in the last few months. It also announced an AI Productivity Guarantee, covering up to $10M in Devin usage if it does not deliver positive ROI. That framing is more interesting than the guarantee itself. Once agents move from experiment to daily engineering habit, the buyer starts asking different questions: How much token spend produced accepted work? How many retries were waste? Which tasks should use a cheaper model? How much human review was saved or added? Which failures created rollback risk? How do you convert time saved into dollars without lying to yourself? The winning coding-agent product has to run, explain, and account for the work. For enterprise agents, observability, evals, cost attribution, and reviewer trust are becoming the same system.
- 27
- 28
- 29
@nichochar ·
> need to build tiny feature (15 LOC) > ask Devin, gets to work > spins up a whole computer to test it with browser use, 500M tokens burned > works, submits to github > 5 other agents (codex, devin, cursor bugbot, etc...) all perform a review > for each comment, each bot polls and responds > small improvement found, Devin fixes it > round 2 of the five agents reviewing > tokens go brrrr > LGTM, merge > 2h, 1B tokens later, tiny feature shipped. is this agi?
- 30
@tristanbob ·
It's been a few months since I last used @DevinAI Desktop (formerly Windsurf). I'm using the "Adaptive" mode because I want my credits to be used as efficiently as possible. (BTW, I'm still using the grandfathered original plan from when Windsurf first became available!) First observation, it produces incredibly verbose thinking output. I love reading thinking about, but this was too much. You can see it repeating things, checking rules, correcting itself, etc. Second observation, it worked great! It pushed two PRs to an open source project after testing them locally. Have you tried Devin recently?
Best Tweets by Topic