0824 | Local Music Organizer for Mac, OpenLogi, Yattayo, and Flown

Mia: Welcome back to ProductHunt Daily on Bri Radio. I'm Mia, and today we've got a packed lineup of new tool launches.

Milo: And I'm Milo. We're looking at a fresh take on the to-do list, plus ways to tidy up your browser tabs and your music library.

Mia: There's also a Rust-based alternative for Logitech mouse software, an official learning hub from Anthropic, and a few more developer-friendly finds.

Milo: Lots to get into. Let's jump in.

Milo: Mac owners with large local music libraries now have a dedicated toolkit that runs entirely offline. Local Music Organizer works natively on macOS, handles Music.app libraries and standalone audio files, and needs no Apple Music subscription. It grew out of the maker's frustration with playlist shuffling and expanded into finding duplicate and missing files, comparing playlists, checking metadata, and detecting suspicious upscales and transcodes. Most analysis is read-only, anything that modifies files needs an explicit action from you, and the product page says music never leaves the Mac, with no analytics, no accounts, and no uploads.

Mia: What I find interesting is the authenticity analyzer. It hunts for technical markers of lossy-to-lossless conversion, suspicious bandwidth limits, bitrate mismatches, and possible sample-rate upscaling — but those are estimates you review yourself, not automatic verdicts, and the dashboard categories work as filters rather than final judgments. It requires macOS 14 or newer on Apple Silicon or Intel, and persistent indexing plus paged results keep large collections practical. It's a free Mac App Store download with full access for three days, the trial doesn't renew and creates no automatic charge, and after that continued use needs a one-time unlock for twenty-nine ninety-nine, with regional prices and taxes set by Apple. No subscription, no sign-in.

Milo: For Logitech Options+ users, there's a new local-first alternative called OpenLogi. Developer AprilNEA built it in Rust and released it under MIT and Apache-2.0, with no affiliation to Logitech. It talks directly to your Logitech devices over HID++ to remap buttons, set DPI presets, toggle SmartShift, and run forty-four built-in actions plus custom shortcuts and scripted actions. It covers Bolt, Unifying, Lightspeed, Bluetooth, and wired USB on macOS, Windows, and Linux, and shows battery level for every paired device. The maker's motivation is visible in the design — they loved the MX Master hardware but found Options+ required an account prompt, a background service, telemetry, and hundreds of megabytes just to change what the thumb button does — so configuration lives in a plain TOML file you can copy between machines, with no account, no telemetry, and update checks off by default.

Mia: The gaps are worth knowing. Per-app profiles aren't available yet and are scheduled for a later release, and the Options+ importer isn't here either, though bindings can be rebuilt as a short TOML file in the meantime. Logitech Flow sits at the far end of the roadmap because a cross-computer clipboard bridge is a very large feature, though Easy-Switch host switching already ships. Bolt pairing works in the GUI, with Unifying and Lightspeed pairing still in progress. On macOS, side-button remapping uses a CGEventTap, which is why the system asks for Accessibility permission.

Milo: Bose QC Ultra owners who keep juggling apps on their Mac now have a menu-bar utility for that. ANCBuddy is independent, not an official Bose product, and it pulls Quiet, Aware, and Immersion mode switching, battery status, and sound profiles into the menu bar. It talks directly to paired QC Ultra devices over Bluetooth, so the Bose Music phone app doesn't need to be running. It supports QC Ultra Headphones Gen 1 and Gen 2 and QC Ultra Earbuds 2nd Gen, and runs on macOS 12 Monterey or newer on Apple Silicon or Intel, signed and notarized.

Mia: The standout feature is AI Auto-EQ, and it's opt-in. When enabled, it reads the current track and applies a visible bass, mid, and treble profile. The developer says it sends the artist, title, and album to ANCBuddy's relay only while that feature is on, caches results by a hashed track key, and doesn't store raw song names. It's a one-time nine ninety-nine after a fourteen-day free trial, with no subscription, updates included, and a fourteen-day no-questions refund through Lemon Squeezy, and version two migrates existing BoseControl license and trial data automatically on first run.

Milo: There's one open question hanging over it. A community member asked whether ANC mode and battery state track correctly when QC Ultra's Bluetooth multipoint switches to another device, or whether the state is scoped only to the Mac — and that still hasn't been answered. It's not compatible with Sony, Sennheiser, or AirPods.

Milo: There's a new to-do app that turns a physical orange slider checklist board into 3D software. Yattayo, from solo developer Ryo Tsudukihashi under the name TsuzuKit, makes every task an actual slider — dragging the knob produces ratchet clicks, ends with a thunk, and drops an OK stamp when the task is done. Each task is a sticky note you can handwrite or type on, swiping a finished note peels it off the board, and you can spin the whole board around with your finger. Home Screen widgets show a real photo of the board, so you can slide a task to done without opening the app.

Mia: It syncs over iCloud across iPhone, Apple Watch, and Mac, needs iOS 17, macOS 14, watchOS 10, or visionOS 1, and ships in English and Japanese. It's free for up to three tasks, then Yattayo Premium is ninety-nine cents weekly or nineteen ninety-nine for life, unlocking unlimited tasks, daily or weekly repeating tasks, and no ads. The maker says it started from a photo of a fridge slider checklist, and they're open on their own question — is physical-feeling UI worth the effort, or just a gimmick? Deadline reminders arrive two days before, the day before, and on the due date, with a gentle daily nudge if a task slips past, and you can mark tasks done straight from the notification. The app hasn't drawn enough ratings or reviews yet to show an overview, so launch reception is still unmeasured.

Mia: There's a new browser extension on Product Hunt called Tab Notes that turns every new tab into a distraction-free notepad. The maker's pitch is that standard note apps feel too heavy and default browser start pages are cluttered, so they wanted a fast, local-first scratchpad sitting right inside the tab. It supports rich text and Markdown with headings, lists, tables, and syntax-highlighted code blocks, plus an integrated canvas for drawing diagrams and flowcharts with connecting arrows.

Milo: It works completely offline on local browser storage, or it can sync to a personal Google Drive — the maker calls that zero server lock-in. You can pull notes up in full-screen tabs, through a quick toolbar popup over any website, or export them as plain text or Markdown. One community member flagged the thing that matters most to them: friction at capture time. If opening a notepad takes more than the half-second it takes to open a new tab, they won't use it, and the new-tab takeover is the thing that addresses it.

Milo: The same commenter raised an open question nobody has answered yet, though — when other extensions also want to override the new-tab page, does Tab Notes detect that conflict and warn you, or does whichever extension loaded last win, leaving you to manually disable the other one? No resolution of that question appears in the discussion.

Mia: There's an app on Product Hunt called Flown built around one private map that draws every flight you've ever taken, tacking on a compensation calculator. It ticks off 197 countries, collects passport stamps, and shows you your year in the air — on-device, with no account. When a late flight means the airline owes you between 220 and 520 pounds, Flown works out the amount and tells you.

Milo: That claim drew a sharp community reaction. One commenter pointed out that the airline-compensation line is the last thing in the description, and it's the only part doing real work in drawing people in. The implication is that it stands out as the compelling hook against everything else the app tracks.

Mia: Aximote, a car-data analytics app built for Android Automotive vehicles, has launched an iPhone companion. The co-founder says the original release in May lived almost entirely inside the car, and today's version closes that gap — trips now appear automatically in a personal dashboard, with no OBD dongle and no manual logging. The makers list a customizable dashboard with more than twenty vehicle metrics, a new Efficiency Score, trip, charging, consumption, and cost analytics, cross-trip trends and comparisons, multi-car support keeping one continuous driving history, and redesigned light and dark interfaces.

Milo: They say history stays with the Aximote account rather than one car or one manufacturer's app. Since launch, more than ten thousand drivers have joined and over a million trips have been tracked, and they say community feedback shaped this release. One commenter called it more of a fitness app for driving than a car app — covering economic, sporty, efficient, and long-haul driving. Another praised the lack of a dongle and cross-brand support, saying most trip-logging apps they'd tried need a plug-in dongle or a single manufacturer's API. One simply wrote, no Teslas.

Mia: The biggest open question is what's underneath it all. Asked whether Aximote pulls from each maker's official connected-car service or from a universal telemetry standard, one commenter said the answer decides which of their family's cars shows up — and that underlying data source remains unanswered.

Mia: FetchSandbox MCP is a new Product Hunt launch aimed at developers using AI coding assistants like Claude Code, Cursor, Cline, Windsurf, and Codex. Co-founder Raj's framing is that AI agents can write integrations — say, one for Stripe — but they can't run them, so code may pass continuous integration while the underlying data stays wrong. The tool's gate asserts the exact end state a correct implementation leaves behind, and it refuses to go green when it can't first reproduce the bug.

Milo: There's an illustrative case worth the detail: a customer paid for five seats, a retry created ten, then fifteen, and an agent's fix stopped the duplicates but left nobody with any seats — yet the tests still passed, because the duplicates were gone. The company describes the output as a receipt, not a vibe. Setup is one block in the MCP config, no API key, no signup, and the catalog has more than seventy ready-made sandboxes for things like Stripe, HubSpot, Clerk, Resend, and Twilio — free to try.

Mia: The community buys the underlying problem but has open questions. One commenter reported that a 200 response with a plausible but wrong body is a failure that survives every check they've tried, and asked whether the proof step asserts the response shape or the actual values. Others asked whether custom internal enterprise APIs can be added to the sandbox list, whether it can test reordered events and delayed retries across the same workflow, whether the proof step can start from a state a bug already wrote rather than a clean sandbox, and whether the receipt catches partial failures. None of those have been answered.

Mia: Anthropic has put its training into one place — Claude Academy, launched on Product Hunt as its official learning hub at academy.claude.com. The maker's pitch frames it as a free, self-paced environment with courses, developer tutorials, and official certification paths, stretching from basic prompting and AI fluency up to complex multi-agent workflows. The launch description also points at agentic-architecture guidance for Claude Code, subagents, and the MCP protocol, plus role-based tracks aimed at developers, creative professionals, and leaders. Those are still maker claims, not verified results.

Milo: What the site's own catalog actually shows is product-specific learning for Claude.ai, Claude Cowork, Claude Code, Claude Tag, and Claude Platform, alongside an AI Fluency section. That section includes a Foundations course built on a four-D framework — Delegation, Description, Discernment, and Diligence — running fourteen lessons with one quiz over about four hours. It also lists a seven-minute tutorial on the four properties of AI, and a separate thirteen-lesson course on AI capabilities and limitations, covering next-token prediction, knowledge, working memory, steerability, and context limits across roughly three and a half hours.

Mia: Hmm. One detail that stands out from community discussion: a Claude Code user described their recurring problem as deciding when to split work across subagents versus running everything in one long context, reporting wasted context from under-splitting and lost coordination time from over-splitting. It's an open question whether the developer track actually addresses that specific trade-off.

Milo: There's a small open-source tool on Product Hunt called Yatko, at yatko.app, that its maker describes as turning any public GitHub repository into clean download links. The pitch is that GitHub Releases can feel complex and intimidating for beginners, forcing users to hunt through the Releases page for the install file that matches their operating system and architecture. The claimed workflow is simple — you enter a repo name on the front page, or swap the domain of any public GitHub URL from github.com to yatko.app, and it picks the right release asset for each visitor's OS.

Mia: The site lays out a few endpoint patterns: yatko.app with owner and repo as the landing page, a direct download path under dl, and a JSON link API under api for CI pipelines and scripts. It even offers cli/cli, neovim/neovim, astral-sh/uv, and BurntSushi/ripgrep as repos to try, with the maker's own repo at github.com/argval/yatko.

Milo: But launch day reports undercut that promise. One commenter said yatko.app/cli/cli and yatko.app/argval/yatko both sat on "Loading release…" forever in Chrome on macOS, with the network panel showing no request to api.github.com at all — which they read as the page hanging before the fetch, not getting rate-limited, since OS and architecture detection is the only code running ahead of that call. Another commenter called it a fix for a tiny GitHub UX problem almost everyone has run into, but wondered how it handles unusual release naming or multiple valid binaries for the same OS and architecture. Neither of those questions has been answered yet.

Mia: So whether it's a one-purchase offline toolkit for repairing a large Mac music library, or an open local-first rewrite of Logitech's options software, neither asks for a subscription.

Milo: Right, and both put the user in control of the data. Thanks for listening, take care.