
Everyone has figured out how to generate code. Almost no one has figured out how to QA it at scale or monitor it in production. @rbren_dev of @OpenHandsDev on agents that start work on their own, and the cloud they run in. do.co/3WqfseS

CEO and co-founder of OpenHands github.com/OpenHands/OpenHan…

Everyone has figured out how to generate code. Almost no one has figured out how to QA it at scale or monitor it in production. @rbren_dev of @OpenHandsDev on agents that start work on their own, and the cloud they run in. do.co/3WqfseS

The same LLM can deliver very different results depending on the harness running it. Tools, context, execution, and verification all matter. That’s the premise behind OpenHands: model choice backed by the infrastructure to put agents to work. Check out the full article from iTech Guides: hubs.la/Q04DW_W30

What if you could train a model for your agent harness instead of adapting your harness to the model? OpenHands + @Snowflake Cortex Training let developers post-train open-weight models using their existing agent configurations and tasks. See how it works 👇 openhands.dev/blog/post-trai…

Huge congrats to @Recursive_SI on the launch!
Excited to support for @Recursive_SI Flint 🚀 40B parameters, just 2.5B active per token, built for fast local coding inference. We’ve tested it with OpenHands. Now developers have another open model for real engineering workflows. openhands.dev/blog/recursive…

OpenHands now integrates with @DigitalOcean Managed Agents! 🌊 Run AI coding agents in isolated cloud environments, pause and resume sessions, and manage everything from Agent Canvas. Use DigitalOcean for agent execution, model inference, or both. More flexibility in where agents run and which models power them. Read more about the integration and how to get started: hubs.la/Q04B3Hbl0

Agentic development isn’t one task for one agent. It’s a system: triage the issue, write the code, run it safely, review the PR, and keep the work moving when feedback comes in. That’s where we focused in September across eight Agent Canvas releases: ⚙️ More powerful automations that carry context from issue → development → PR review 🎛️ Agent Profiles with scoped tools and credentials, so each agent can be configured for the job it needs to do 🐳 Isolated Docker runtimes that give each conversation its own execution environment We also made Apps easier to build and share, alongside dozens of improvements across the Agent Canvas workspace. Read the full September product update: hubs.la/Q04zv2nW0

"If you commit to one particular lab, there's a good chance you're going to be left behind." @OpenHandsDev CEO @rbren_dev on choice and owning your intelligence. Hear more from him at the Open Intelligence Summit, Oct 13 in SF: do.co/4ArVTlL

90K GitHub stars ⭐ More than a number, it’s a reflection of the community building OpenHands alongside us. Every contribution, PR, issue, experiment, and piece of feedback makes the project better. Thank you to everyone building with us in the open!

Today we’re launching micro1’s PII transformation model, flow-transform 1.0, delivering frontier-level performance across detection, identity synthesis, and transformation of personally identifiable information. On PrivacyBench, our model reaches 96.0% F1, outperforming every detection baseline we tested, including Tonic Textual, Claude Opus 4.8, Sonnet 4.6, Microsoft Presidio, Haiku 4.5 and GLiNER2. Some of the most valuable training data for frontier AI models lives inside fully functioning companies. It captures years of real work across decisions, communications, tools, handoffs, exceptions and the relationships connecting them. The problem is that this data is also full of PII. Traditional redaction makes the data safe, but it also destroys the very workflows and relationships frontier models need to learn from. flow-transform 1.0 solves this by turning enterprise operational data into high-fidelity training data for frontier models by replacing real-world identities without flattening the reality the data captures.

Excited to be welcoming @rbren_dev, @ivanburazin, @alxnails, and Leix Xu to the Open Intelligence Summit. Open intelligence is more than just open weights, it's providing builders with choice and ownership at every layer of the stack, agents to infra. Join us!
The first speakers for Open Intelligence Summit are live. If you're building an AI-native company, the agenda covers the decisions in front of you: who owns your data, what inference costs you, how your agents run, and what your users experience. Oct 13, SF. Request a seat: do.co/4ArVTlL

We now run OpenHands in our own cloud. The agent implements a task, reviews its pull request, and repeats review and fixes before handing it to a human. We still decide whether to merge. hajek.no/posts/2026/shared-a…

"openhands...the autonomous coding agent has now doubled for four straight months" 🦾
The August ClickPy report is out: overall downloads cooled over summer time, PyPI recorded 176.6B downloads in August 2026, down 4.5% from July. But MCP and AI-agent tooling are still accelerating. 🤖 Notable packages this month: 🔌 mcp-types: 11.7M → 61.3M (+425%), MCP tooling just became a 60M-download dependency ☁️ google-cloud-parametermanager: 862K → 4.82M (+459%), @googlecloud, a 5.6x jump for Google Cloud’s Parameter Manager client 🎨 skia-python: 167K → 4.02M (+2,312%), Python bindings for Google’s graphics engine suddenly went 24x 🤝 openhands: 1.11M → 2.52M (+128%), @OpenHandsDev, the autonomous coding agent has now doubled for four straight months 📈 metricflow: 4.31M → 8.11M (+88%), @dbt_labs, semantic-layer adoption keeps compounding Full report here: buff.ly/fscHSko


Really enjoyed this conversation w/ Laura at Notable!
A developer prompting their way through a Jira ticket gets about 20 to 30% faster, says @rbren_dev. So that's why he says the real gains sit somewhere else. On the new episode of First Commit, Laura Hamilton sits down with the co-founder and CEO of @OpenHandsDev, the open source project that became one of the most starred AI coding agent repos on GitHub.

Super happy to see Nemotron continuing to evolve. OSS models are becoming the norm!
We’re proud to be an @NVIDIAAI launch partner and excited to support NVIDIA Nemotron 3.5 Lightning! ⚡ Always-on agents need fast, customizable models, but they also need infrastructure: harnesses, runtimes, sandboxes, integrations, policies, audit trails, and deployment flexibility. That’s where OpenHands comes in. Learn how to use Nemotron 3.5 Lightning & OpenHands: openhands.dev/blog/nvidia-ne…

🦾
OpenHands crushed Codex by 2.3× on token efficiency! We gave both agents the same task on the same model (Qwen3.5 35B): build an 8-bit Space Invaders in 3 iterations (build, fix, polish). Output: • OpenHands: 219K tokens, • Codex: 513K tokens, @OpenHandsDev beats Codex in local running. The difference is in how they handle context. OpenHands reuses the unchanged data across every pass and only pays for new tokens, while Codex re-sends and re-counts it every iteration. OpenHands finished a bit slower, but on local runs, tokens spent are the real cost.