{"slug": "meridian-might-be-the-first-local-first-tool-that-actually", "title": "Meridian might be the first local-first tool that actually", "summary": "Meridian, a local-first productivity tool that uses OCR and a local LLM to generate work logs, reached #1 on Product Hunt today. The tool, which runs locally with an encrypted SQLite store and supports Ollama by default, generated standup drafts that were 90% ready to paste during a week of testing. Setup takes about 15 minutes on Fedora 39, and the MIT-licensed project allows forking to create a CLI/TUI version.", "body_md": "# Meridian might be the first local-first tool that actually\n\n*that*you were in VS Code or Chrome, but not\n\n*what*you were actually doing. Meridian takes a fundamentally different approach. It runs locally, watches your screen activity via OCR and window titles, then uses a local LLM (Ollama by default, though you can point it at any OpenAI-compatible endpoint) to reconstruct narrative summaries.\n\nThe architecture is surprisingly clean. A lightweight Electron wrapper spawns a Rust sidecar that handles the encrypted SQLite store — AES-256-GCM, keys never leave the machine. The capture loop runs every 30 seconds by default, pulling active window title, process name, and a screenshot region. That raw stream gets batched and fed to the LLM with a prompt that essentially says \"turn this chronological noise into a coherent work log.\" The output lands as draft entries in a local web UI where you can edit, tag, and push to Jira, Linear, GitHub, or Azure DevOps.\n\nWhat impressed me most during a week of daily driving it: the prompt engineering is *good*. It correctly distinguished between \"debugging the auth middleware\" and \"reading Stack Overflow about JWT refresh tokens\" — same files, totally different intent. It also caught context switches I'd forgotten: that 20-minute Slack thread about database indexing that turned into a real schema change. The daily standup draft it generated was 90% ready to paste.\n\nA few rough edges worth knowing. The Electron UI feels a bit heavy on memory (~400MB idle). The OCR occasionally garbles terminal output with weird fonts — I ended up adding a regex filter in `config.toml`\n\nto skip windows matching `^alacritty.*$`\n\nand just rely on the shell history integration instead. Speaking of which, the shell history hook (zsh/fish/bash) is a separate binary you install via `cargo install meridian-shell`\n\n— not documented prominently enough. Without it, terminal work shows up as generic \"terminal activity.\"\n\nSetup took me about 15 minutes on Fedora 39:\n\n```\n# 1. Grab the AppImage from releases\nwget https://github.com/meridian-dev/meridian/releases/download/v0.4.2/meridian-0.4.2.AppImage\nchmod +x meridian-0.4.2.AppImage\n\n# 2. Install Ollama if you don't have it\ncurl -fsSL https://ollama.com/install.sh | sh\nollama pull llama3.1:8b\n\n# 3. Run once to generate config\n./meridian-0.4.2.AppImage --init\n\n# 4. Edit ~/.config/meridian/config.toml — point llm_endpoint at http://localhost:11434/v1\n#    and set model = \"llama3.1:8b\"\n\n# 5. Optional but recommended: shell history integration\ncargo install meridian-shell\nmeridian-shell install --shell zsh\n\n# 6. Launch\n./meridian-0.4.2.AppImage\n```\n\nThe MIT license means you can fork and strip the Electron layer entirely if you want a pure CLI/TUI — someone's already started a `meridian-tui`\n\nexperiment in the discussions tab.\n\nFor teams drowning in status-update theater, this hits a sweet spot: zero SaaS dependency, auditable local data, and output that actually resembles human communication. The fact that it's #1 on Product Hunt today suggests I'm not the only one tired of performing productivity for Jira.\n\n[Agile development is basically just a fragmented waterfall now 4d ago](/en/news/6548/)\n\n[Stop tracking \"AI adoption rates\" because they are essentially 10d ago](/en/news/5722/)\n\n[Next Huzzah flips the agent loop: you write pseudocode →](/en/news/7087/)", "url": "https://wpnews.pro/news/meridian-might-be-the-first-local-first-tool-that-actually", "canonical_source": "https://promptcube3.com/en/news/7089/", "published_at": "2026-08-20 20:03:20+00:00", "updated_at": "2026-08-20 20:14:57.839292+00:00", "lang": "en", "topics": ["ai-tools", "artificial-intelligence", "large-language-models"], "entities": ["Meridian", "Ollama", "Electron", "Rust", "Jira", "Linear", "GitHub", "Azure DevOps"], "alternates": {"html": "https://wpnews.pro/news/meridian-might-be-the-first-local-first-tool-that-actually", "markdown": "https://wpnews.pro/news/meridian-might-be-the-first-local-first-tool-that-actually.md", "text": "https://wpnews.pro/news/meridian-might-be-the-first-local-first-tool-that-actually.txt", "jsonld": "https://wpnews.pro/news/meridian-might-be-the-first-local-first-tool-that-actually.jsonld"}}