Aside hears your meeting, transcribes it on your computer, and streams AI answers into a floating panel beside your call — then writes up what was said when you hang up.
Audio stays on your computer · Free 7-day trial · Cancel anytime
Find us on Product HuntWorks anywhere your computer can hear — Zoom · Google Meet · Microsoft Teams · Webex · phone bridges
A full meeting copilot that lives in your menu bar or system tray — for sales calls, support, standups, client meetings, interviews, and lectures.
A local Whisper model turns speech to text entirely on-device, with three speed / accuracy tiers. Your audio isn’t uploaded — the one exception is cloud transcription, off by default, for PCs that can’t keep up on their own.
Aside detects questions in real time — in English and Spanish — and streams a Claude-powered answer instantly. Three models: Fast, Smart, and Pro.
The panel floats beside your call and never steals keyboard focus — you keep typing in whatever app you were already in. Drag it anywhere, resize it, or send it away with a hotkey.
Aside captures your mic and your system audio separately, tagging every line You or Them so answers land with the right context.
One hotkey captures your screen and answers what’s on it — a dense spreadsheet, an error message, a contract clause, a form you’ve never seen before.
Press ⌥⌘A to ask anything without leaving your call. Nudge any answer Shorter, Deeper, with an Example, or Simpler.
Every session is saved with an AI wrap-up: overview, key questions, action items, and a follow-up email draft ready to send.
Feed it your résumé or notes and transcription nails names, products, and acronyms. Answer style presets and configurable hotkeys make it yours.
No bots joining your call, no browser extensions, no setup ritual. Aside works from the menu bar — or the system tray on Windows.
Zoom, Meet, Teams, Webex, a phone bridge — if your computer can hear it, Aside works with it. It even offers a gentle nudge when your meeting app opens.
A local Whisper model transcribes in real time and tags each line You or Them. By default nothing is uploaded and nothing leaves the machine.
Detected questions stream Claude-powered answers into a small floating glass panel. Nudge any answer shorter, deeper, or simpler without breaking your flow — and get a written wrap-up when the call ends.
Transcription runs on-device with a local Whisper model. Aside doesn’t record to the cloud, doesn’t upload your audio, and doesn’t store your conversations on our servers.
When you ask for an answer, only the essentials — the question and a short transcript excerpt, or your screen capture — are sent to generate it. That’s the whole exchange.
There is one exception, and it’s yours to make. A PC with no GPU acceleration can run the local models slower than real time, which makes Aside drop parts of the conversation. If that happens to you, Settings offers cloud transcription instead. It is off by default, Aside states plainly what changes before you turn it on, and audio sent that way is transcribed and discarded — never stored, never reviewed.
Read the privacy policy →Seven days and 40 answers on us — no card required.
Kick the tires on a real call.
For regular meetings that matter.
$15.83/mo, billed annually
Maximum firepower for high-stakes calls.
$32.50/mo, billed annually
Answer caps reset monthly on every plan. Cancel anytime — you keep access until the end of your period. Payments handled securely by Paddle.
Fittingly.
One answer is one AI response: a detected-question answer, an Ask-anything reply, a Screen Assist result, or a refinement like Shorter or Deeper. On-device transcription is unlimited and never counts against your cap.
Effectively all of them. Aside listens to your computer’s microphone and system audio, so Zoom, Google Meet, Microsoft Teams, Webex, and even phone bridges work out of the box — no plugins, no bots joining your call. If your computer can hear it, Aside can help.
Not unless you choose to. Out of the box, transcription happens on your computer with a local Whisper model and your audio never leaves it. When Aside generates an answer, it sends only the detected question and a short transcript excerpt — or your screen capture, for Screen Assist — to our API, which forwards it to Anthropic. We don’t store any conversation content, just usage counts for billing.
The one way audio leaves your machine is cloud transcription, a setting you turn on yourself. It is off by default and exists for PCs without the GPU power to transcribe in real time — on those machines the local models fall behind live speech and Aside starts skipping what was said. Turn it on and each utterance is transcribed on our servers instead, then discarded; the on-device model stays loaded and takes back over if a request fails.
No. The panel is excluded from screen capture by macOS and Windows — the same mechanism password managers use to keep credentials out of recordings. It’s there so your own notes and half-written drafts don’t end up in a recorded demo, the way you wouldn’t want your unread messages in one either.
That’s a privacy feature, not a disguise. Aside isn’t built to hide that you use it, and we’d tell anyone to mention it the same way they’d mention taking notes. Where you are asked not to use an assistant — a proctored exam, an assessment with its own rules — don’t.
Answers pause until your cap resets on your billing date — or resume immediately if you upgrade. Transcription keeps running on-device either way.
Yes. Manage or cancel your subscription through the Paddle billing portal in a couple of clicks. You keep full access until the end of your current billing period.
On a Mac: macOS 14.6 or later; Apple Silicon is recommended for the best on-device transcription. On first launch you’ll grant microphone and system-audio permissions so Aside can hear both sides of the call.
On Windows: Windows 10 (version 2004) or Windows 11, 64-bit. A GPU with Vulkan drivers (any recent NVIDIA, AMD, or Intel) keeps transcription real-time; without one it runs on the CPU, just more slowly. The download is a ZIP you unzip and run — no installer, no admin rights, everything bundled.
Either way, plan on 8 GB of RAM and about 1 GB of free space for the app and its on-device speech model. Answers need an internet connection; transcription never does.
Seven days free. Forty answers. No card. If it doesn’t make your next meeting easier, delete it.
v1.0.2 · DMG, 4.3 MB · Notarized by Apple · Free 7-day trial
macOS 14.6+ · Apple Silicon recommended
v1.0.1 · 64-bit ZIP, 96 MB · Unzip and run — no installer · Free 7-day trial
Windows 10 (version 2004) or Windows 11 · GPU with Vulkan drivers recommended
