{"slug": "why-we-built-sahkarmi-the-open-source-ai-coworker-that-never-acts-without-your", "title": "Why We Built Sahkarmi: The Open-Source AI Coworker That Never Acts Without Your Approval", "summary": "Dhrubojyoti Gangopadhyay, founder of Sahkarmi, released an open-source (Apache-2.0) desktop workspace in which autonomous AI coworkers run locally and pause for human approval before any sensitive action such as shell commands, file writes, web form submissions or transactions. The Windows 10/11 x64 build ships now with a sandboxed terminal, filesystem access, an embedded Chromium browser and persistent memory, while macOS and Linux installers exist in the repo but are not released. Gangopadhyay said he wants feedback on where the approvals model gets in the way: \"that feedback is what shapes the roadmap.", "body_md": "**By Dhrubojyoti Gangopadhyay (Founder, Sahkarmi)**\n\n*Website: [sahkarmi.site](https://sahkarmi.site) | GitHub: [github.com/ganguydhrubo/sahakarmi](https://github.com/ganguydhrubo/sahakarmi) | License: Apache-2.0*\n\nOver the past two years, the promise of autonomous AI agents has captured the imagination of every software engineer and startup founder. We were told agents would write our code, debug production incidents, handle customer support, and deploy our cloud infrastructure while we sleep.\n\nYet in practice, most developers still hesitate to hand full control over to autonomous scripts. Why?\n\n`rm -rf`, drop a database table during a migration attempt, or overwrite critical configuration files.\nWe asked ourselves: **What if autonomous AI agents didn't replace human judgement, but instead operated like trusted colleagues with built-in human verification gates?**\n\nThat is why we built **Sahkarmi (सहकर्मी)** — the Hindi word for *coworker*.\n\nSahkarmi is an **open-source desktop team workspace** where autonomous AI coworkers run locally on your own machine.\n\nUnlike traditional scripts or chat windows, Sahkarmi's agents are equipped with specialized tools — a sandboxed local terminal, filesystem access, an embedded Chromium browser, and persistent long-term memory — while strictly obeying a single foundational principle:\n\n**Zero Unapproved Impact**: An agent can plan, inspect files, draft code, and browse documentation autonomously; but the moment it attempts a sensitive action — such as executing a shell command, writing a file, submitting a web form, or executing a transaction — it pauses and presents a clear, interactive **Human-in-the-Loop Approval Card**.\n\n```\n[ Sahkarmi Human Verification Card ]\n--------------------------------------------------\nTask: Deploy backend schema migration\nAgent: Dev_Backend_Coworker\nProposed Command: python manage.py migrate --database=prod\nWorking Dir: /home/user/project\n\n[ APPROVE ]     [ REJECT ]     [ REVISE ]\n--------------------------------------------------\n```\n\nYou maintain complete oversight. You can approve with a single keystroke, reject with corrective feedback, or take over directly.\n\nSahkarmi never locks you into a proprietary LLM API. You can switch models per task:\n\nSahkarmi is free and open-source under the **Apache-2.0** license.\n\n**Platform support, honestly stated:** Sahkarmi ships a **Windows 10/11 (x64)** desktop build today. macOS and Linux are planned but are *not* released yet — installer scripts exist in the repository but are not used for releases.\n\nGrab the latest signed release directly:\n\n`SHA256SUMS-windows.txt` so you can verify the download before running it.\n\n```\ngit clone https://github.com/ganguydhrubo/sahakarmi.git\ncd sahakarmi\nbun install\nbun run dev\n```\n\nWinGet and Scoop manifests are submitted and currently awaiting review by their respective communities. Once those land, installation will also be a one-liner. Until then, please use the direct download above.\n\nWe believe the future of software development belongs to collaborative human-AI teams, where humans provide intent and strategic oversight while autonomous coworkers handle the execution grind.\n\nIf you try it, I would genuinely like to hear where the approvals model gets in your way — that feedback is what shapes the roadmap.", "url": "https://wpnews.pro/news/why-we-built-sahkarmi-the-open-source-ai-coworker-that-never-acts-without-your", "canonical_source": "https://dev.to/ganguydhrubo/why-we-built-sahkarmi-the-open-source-ai-coworker-that-never-acts-without-your-approval-51mo", "published_at": "2026-10-27 10:33:13+00:00", "updated_at": "2026-10-27 10:42:21.217068+00:00", "lang": "en", "topics": ["ai-agents", "ai-tools", "developer-tools", "ai-safety", "artificial-intelligence"], "entities": ["Sahkarmi", "Dhrubojyoti Gangopadhyay", "GitHub", "Apache-2.0", "Windows", "Chromium", "WinGet", "Scoop"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/why-we-built-sahkarmi-the-open-source-ai-coworker-that-never-acts-without-your", "markdown": "https://wpnews.pro/news/why-we-built-sahkarmi-the-open-source-ai-coworker-that-never-acts-without-your.md", "text": "https://wpnews.pro/news/why-we-built-sahkarmi-the-open-source-ai-coworker-that-never-acts-without-your.txt", "jsonld": "https://wpnews.pro/news/why-we-built-sahkarmi-the-open-source-ai-coworker-that-never-acts-without-your.jsonld"}}