Your agent is smart. Its tools should be too. Check composio.dev
build on top of us ➡️
- Tweets 2,001
- Following 49
- Followers 24,564
- Likes 2,036
Composio retweeted
How time flies. I've been @Composio 8 months now building out our content engine and podcast: interviewing founders about what they're building.
Season 3 starts soon and it'll be different.
Every episode is one company, one department, filmed inside. Less "what do you think about AI," more what does it actually do here daily.
I'll be asking questions like:
-How much of your product is built by your own agent
-What you automated and had to take back
-What roles you stopped hiring for
-What you'd do first at a slower company
If you're a company in San Francisco (or Toronto;)) and you'd let a camera in, DM ME!
Episode 1 is a works in progress right now;)
5
2
45
1,709
We tested 6 AI models on 30 challenging agent tasks: GPT-6 Astra, Opus 5.5, GPT-6 Sol, Pareto 26.9, DeepSeek V4 Pro, and GLM 5.3 Flash.
Sol matched Opus’s score, finished faster, and cost about a quarter as much per successful task.
Here’s how all 6 models compared 🧵🧵🧵
52
26
335
142,405
Composio retweeted
Spend your development time on the things that differentiate your agent from the rest of the market.
Let Composio handle the rest - connect your agent to over 1,500 apps for your users with one integration to @composio.
Auth, tool calls, sandboxing, and more with one integration to Composio.
2
2
10
878
Composio retweeted
Atomic Agent connects 1500+ apps powered by @composio
✦ Set up with one API key
✦ Connect Gmail, Slack, Notion, GitHub and more
✦ Send emails, create tickets and manage files
Available on Mac, Windows and Linux: github.com/AtomicBot-ai/atom…
11
7
38
2,898
We aren't even close to running out of ways to use Composio at this point...
There's only one classifier that matters: what Mean Girls character are you based on your HackerNews comments.
With @typesafeai and @ComposioDevs we can finally answer that question (in a type-safe way).
1
16
2,965
Here's how all 6 harnesses compared on task success rate:
Codex, Hermes Agent, and Command Code tied for the lead at 21/29. Claude Code was one task behind, while OpenCode and Pi Agent finished at 19/29.
The spread was about 7 percentage points. Only 3 tasks produced different results across harnesses: 18 passed across the board, and 8 tasks also failed across all harnesses.
6
8
108
82,559
