Social Wire

Page 15 of 16 · newest first.

@_markfenner · Desktop
Traveling today, so we're knocking the work out early. Testers already found a couple bugs in DevinX. Fixes going out before I'm on the road. Happy Thursday 🫡 Worth tracking on the Devin Central desk.
@modelsagree · Cloud
Cognition ranked second in a list of top background coding agents, behind GitHub and ahead of OpenAI
@YouTube · Cognition
A counter-narrative to the "fully autonomous" framing: the most effective agent workflows still depend on human judgment at the review seams. This aligns with what Cognition's own docs emphasize — explicit done-when criteria, easy-to-verify tasks, and human review before merge.
@cognition · Cognition
SWE-1.7 is free for paid Devin users for a month, with Lightning at ~1000 tok/s on Cerebras. If you’ve been curious about Cognition’s own model lane, this is the window — desk write-up:
@Naseem9brev · Cognition
First Cognition meetup in Dubai packed the room on a cold start — 250 signups, founders and YC folks, credits on the table, more mixers coming. Community density outside SF is the story; Ambassadors are turning tip energy into rooms.
@_markfenner · Desktop
The hard part of pocket Devin wasn’t Cloud — it was local Mac sessions with no public API. DevinX Connector + Tailscale + per-request device verify keeps credentials on the Mac while the phone steers. Cloud and local, same mission-control habit.
@cognition · Cloud
From the Wire (@cognition): As of today, Devin Fusion now incorporates Fable 5. Surprisingly, Fable 5 runs at a lower cost per task than Opus 4.8. Even though it’s a more expensive model, we observed cost efficiency improvements in areas like… Source: https://x.com/cognition/status/2076714965344342382
@chrisbarber · Cognition
From the Wire (@chrisbarber): I asked people which companies have the highest density of talented people they know: 1st. Cognition - 9 votes Equal 1st. Anthropic - 9 votes 2nd. Modal - 5 votes 3rd. OpenAI - 4 votes 4th. Standard Intelligence - 3… Source: https://x.com/chrisbarber/status/2075626354406785195
@cognition · Cognition
GPT-5.6 landed in Devin with Cognition highlighting Sol’s cost/perf on their FontierCode numbers. Another reminder the product is a harness — model menu keeps moving; your workflow shouldn’t be glued to one vendor week-to-week.
@dabit3 · Cloud
Nader’s framing of Cloud QA: parallel “human-grade” UI testing — stand up the env, real auth, mouse-and-keyboard through the app, annotated video as proof, skills reused next run. Same tape builders praise when they say “send me the recap.”
@cognition · Cloud
Cognition flipped Devin Fusion to an agent preview in Cloud — the multi-model / manager-style lane they’ve been teasing next to Ultra. Worth pairing with the Fable-cheaper-than-Opus notes if you’re picking defaults for long tasks.
@_markfenner · Cognition
From the Wire (@_markfenner): If we're being completely honest, complaining that any frontier model is "terrible" right now is a skill issue. FOH The Fable vs Sol bullshit is truly preference dressed up as analysis. Fable wins SWE-Bench Pro. Sol… Source: https://x.com/_markfenner/status/2076796803672863059
@_markfenner · Desktop
From the Wire (@_markfenner): Terrible weather this morning, so the run moved inside. Treadmill room was absolutely atrocious, but we got it done anyway. Easy 4 🏃🏾‍♂️. Health is wealth. Yes, I know I need a haircut. Shhh. I'll handle it. Side… Source: https://x.com/_markfenner/status/2076656088737276055
@walden_yan · Cloud
From the Wire (@walden_yan): I remember when @joon_h_lee showed that Fable is cheaper than Opus in our mixed harness. We assumed it must be a bug. But then I tried it, and on one of my tasks, Fusion w/ Fable was 4x cheaper and gave a better PR.… Source: https://x.com/walden_yan/status/2076715678242124154
@nikitabier · Cognition
From the Wire (@nikitabier): We're rolling out a small tweak to boost visibility of your posts to your mutuals (people who you follow back). We noticed this data was missing from the algo and it made your friends appear less in your replies.… Source: https://x.com/nikitabier/status/2076747704248758617
@_markfenner · Desktop
DevinX isn’t just chat-from-phone anymore: Cloud and local sources in one list, plus read-only visibility into security-scan sessions (counts, severity, linked PR). Mobile as mission control, Desktop/Cloud as the launch surfaces.
@ishaangtwt · CLI
A user achieved high productivity with Devin, generating over 80k lines of code and building multiple projects, including a Slack-native autonomous agent.
@o_kwasniewski · Cognition
Integrating Devin with TesterArmy can significantly enhance testing efficiency and confidence in code changes.
@ScottWu46 · CLI
From the Wire (@ScottWu46): One year ago today, Cognition acquired Windsurf, capping off one of the crazier weekends in our company's history. The outline has been told many times: the first call on Friday evening, the rush through Saturday and… Source: https://x.com/ScottWu46/status/2077078038744109398
@Builderio · Cognition
Best short take we’ve seen: Cursor keeps you thinking *through* the code; Devin asks you to hand work *off*. Pick tools by attention model—flow vs relief—not by who wins a feature matrix. Most teams will keep both lanes open: IDE for exploratory edits, agent sessions for scoped overnight tasks.
@TechwithRKM · Desktop
Video desk brief if you missed the Desktop rename: Cascade’s July runway is the part that bites operators, not the new wordmark. Plan Local/CLI-shaped workflows *before* that window closes, not the morning after. Full desk coverage:
@willebrew · Cloud
Builders keep calling out Cloud’s autonomous test + video recap loop: you don’t have to babysit the browser; you get a replay of what Devin actually clicked. Same thread as “record a demo for me” — proof artifacts, not vibes.
@cognition · Cloud
From the Wire (@cognition): Ship straight from the conversation. Devin in @Slack lets your team investigate issues, answer codebase questions, and start dev tasks without leaving the channel. Thanks for the feature, @Slack👇️ Source: https://x.com/cognition/status/2077484850244648998
@dabit3 · CLI
From the Wire (@dabit3): Since you don't need a frontier model typing every line of code, we can optimize subagents beyond just cost. /lightning skill offloads implementation to 1,000 tok/s subagents, resulting in ~5x faster execution. it… Source: https://x.com/dabit3/status/2077456437652119693