Before you begin

System requirements

Friday is designed to run on virtually any modern Windows PC. Check these before starting.

Windows 10 or 11

64-bit builds are recommended, the bundled speech and voice models run noticeably better on modern hardware.

Free disk space

The Installation of Friday needs atleast 1.5GB of free space (Friday files + Ollama).

Microphone

Any USB, built-in, or Bluetooth mic. Set it as the default recording device in Windows Sound Settings.

Audio output

Speakers or headphones required to hear Friday's voice responses.

Speech & voice run offline
Internet only for optional features
Microphone permission required

Step 01

Download the installer

Grab the latest official Friday release directly from GitHub Releases and the Download Page.

Friday-Setup.exe

Download the Friday Installer by using the button below or visiting the download page.

Download Friday
01

Finding the right file on GitHub

Download

Click the button above or go to the GitHub Releases page or the Download Page. Under the latest release, look for the Assets section and click Friday-Setup.exe to download it.

Always download Friday from the official GitHub Releases page and My Website. Third-party mirrors or file-sharing sites may serve outdated or tampered builds.
Browser warning: Your browser may show "This file type may be harmful." This is standard for any unsigned installer. Click Keep or Download anyway, Friday contains no malware. See Troubleshooting if Windows Defender or antivirus flags it.

Step 02

Run the installer

Double-click the downloaded .exe and follow the short Setup wizard.

02

Installing Friday

Install

Double-click Friday-Setup.exe. The Setup wizard unpacks the frozen application files and Ollama Offline model into a folder of your choosing, and creates Start Menu shortcuts.

Windows SmartScreen: You may see a blue dialog, "Windows protected your PC." This appears because the installer isn't signed with a commercial certificate. Click "More info""Run anyway". You'll only see this once.
Recommended install path: Install to C:\Users\YourName\AppData\Local\Friday or C:\Users\YourName\Documents\Friday. Avoid C:\Program Files unless you run as admin, Friday's runtime files (config, TTS cache) live alongside the executable.

Once the wizard finishes, Friday is ready, no additional setup, no dependencies to install, no configuration files to edit by hand.

02b

What's inside the installation

Info

The installer unpacks everything Friday needs to run:

Friday.exe, the frozen application itself.

Moonshine speech model, the offline recognition engine that hears your wake word, commands, and dictation entirely on-device.

Piper voice model + TTS cache, the offline text-to-speech engine Friday speaks with, plus a cache folder that fills up over time for instant playback of repeated phrases.

Config & secrets files, written to %APPDATA%\Friday on first run, and DPAPI-encrypted so they're unreadable outside your own Windows login.

Step 03

Launch Friday

Open the installed folder and double-click Friday.exe to fire it up for the first time.

03

Running Friday.exe

Launch

Open your Friday install folder and double-click Friday.exe, or use the Start Menu shortcut. On first launch it may take a few extra seconds while Friday warms up.

Microphone permission: Windows may ask "Do you want to allow Friday to access your microphone?" Click Yes , this is essential for voice recognition. You can manage this later under Settings → Privacy → Microphone.
The first run takes some seconds because friday os busy setting up everything needed, so WAIT until friday finishes and You will be greeted.
03b

What happens on startup

Info

When Friday starts successfully it will:

1
Greet you with a small briefing containing your system info, headlines, et-cetera, Friday speaks a welcome message using a natural-sounding voice containing your system Info, headlines, reports and others.
2
Begin listening immediately, the microphone activates and Friday starts listening for a wake word. Friday is ready right after the greeting.
3
Load the TTS cache, any previously generated audio is available for instant playback. On first run the cache is empty, so the first few responses generate fresh audio, this is normal.
4
Scan installed applications, Friday builds a catalogue of everything installed on your PC so it can open (or close) apps by name.

Step 04

Set Friday to auto-start

Add Friday to your Windows startup folder so she launches automatically every time you log in.

Don't skip this step

Installing Friday only puts the files on disk, it does not make her launch on its own. If you want Friday ready the moment your PC boots, her shortcut needs to be sitting inside the Windows startup folder.

04

Adding Friday to Windows startup

Auto-start

Follow these steps exactly:

1
Navigate to your Friday install folder, right-click Friday.exeSend to → Desktop (create shortcut). This creates a shortcut on your desktop.
2
Press Windows key + R to open the Run dialog.
3
Type shell:startup and press Enter. A File Explorer window opens showing your personal startup folder.
4
Copy the Friday shortcut from your desktop and paste it into the startup folder that just opened. You can also say open the startup folder once Friday is already running, since start up is one of the folder shortcuts she recognises.
5
Done. Friday will now launch automatically every time you log into Windows, no manual opening required.
Reminder: a fresh install of Friday will not auto-start unless you complete this step. The shortcut is what makes it happen, the install folder alone does nothing at boot.
The startup folder path is %APPDATA%\Microsoft\Windows\Start Menu\Programs\Startup. Only your shortcut goes here, never move the actual Friday install folder. The shortcut is just a pointer; Friday stays where you installed her.
If you ever want to stop Friday from launching on startup, simply delete the shortcut from the shell:startup folder. The Friday installation itself is untouched.

Step 05

Start talking to Friday

Say a wake word, then just speak naturally, Friday listens, understands, and responds.

05

Speaking clearly

Use it

Friday uses Moonshine for fully offline speech recognition, the same engine handles the wake word, spoken commands, and free-form dictation. It performs best when you speak clearly at a moderate pace. You don't need to yell or over-enunciate.

Background noise: Moonshine handles mild ambient noise well, but loud music, TV, or echo-heavy rooms reduce accuracy. A headset microphone gives the best results by far.

All audio is processed entirely on-device by your CPU, no round trip to the internet, so recognition is fast regardless of your connection.

05b

How the Piper TTS cache works

Performance

Every time Friday speaks a new phrase, the generated audio is saved in a local cache folder inside the Appdata directory. The next time she needs that exact phrase, it plays back instantly instead of generating it again.

The TTS cache is never cleared automatically, it only grows, and Friday only gets faster with use. Greetings, time responses, and frequently spoken phrases snap back instantly after their first generation.

Make her yours

Customising every command in the Settings GUI

Almost nothing you say to Friday is hardcoded, every wake word and command phrase shown in this guide can be edited, extended, or removed.

Opening the Settings interface

Customise

Say Configuration, configure, or open config at any time to open the on-screen FRIDAY | SYSTEM INTERFACE window. Every tab holds a set of editable phrase lists, add a phrase, remove one, or add your own alternate wording for how you naturally speak.

Before saving a new phrase, Friday checks it against every other command list in the app. If the phrase is already claimed elsewhere, she'll refuse it and tell you exactly which list is using it, so two commands can never silently race each other.

The tabs below map directly onto the command groups in this guide, open the matching tab to edit any command you see in the explorer further down the page:

GENERAL
VOICE
MEDIA
WINDOWS
MOUSE
SYSTEM
APPS & WEB
TIMERS
WEATHER
NEWS

A few extra tabs round out the rest of the interface: MODES for building your own custom activation phrases and routines, FOLDERS and REMINDERS for the paths and health nudges Friday uses, TEXT TO SPEECH for the Friday's voice itself, plus CONVERTER, DOWNLOADER, CONVERSATION, and CHAT for everything else.

Even the wake word itself is editable, from the GENERAL tab, if "Friday" collides with something else in your daily conversation, add an alternative there.

Voice commands

What you can say

Every command Friday ships with, grouped exactly the way they're grouped in the Settings GUI. Browse the strip below, or let it play through each group on its own.

Phrases shown are the primary way to say each command, most also recognise a handful of alternate phrasings and common misheard variants. Say settings to see (and edit) every variant yourself.

Tips & best practices

Getting the most out of Friday

Small habits that make a big difference in day-to-day use.

Add Friday to shell:startup

See Step 4 above. Placing the Friday shortcut in your Windows startup folder means she's always ready the moment you log in, no manual launching required. This is the single most-missed step during setup.

Open the Settings GUI and make the commands your own

Say "Configuration" and spend two minutes adding the phrasing you'd naturally use. Friday's collision check means you can't accidentally break another command while doing it.

Use a headset or close-field mic

Friday's accuracy improves significantly with a microphone closer to your mouth. Even a budget USB headset beats a far-field laptop mic in a noisy environment.

Set your mic as the Windows default

Go to Settings → System → Sound → Input and confirm your intended microphone is set as the default recording device.

Never move just the .exe

The TTS cache and config files live inside the Friday Appdata folder. Moving only Friday.exe breaks the lookup. If you need to relocate Friday, move the entire install folder and update your shortcut path.

Wait for Friday to finish speaking

Friday processes one command at a time. Issuing another command while she's mid-speech may cause it to be missed. Wait for the response to finish.

Say "rescan" after installing a new app

The app catalogue is built at startup and doesn't update automatically while Friday is running. Say "rescan" or "rebuild" to refresh it after installing something new.

Know the difference between "clear memory" and "reset to defaults"

"clear memory" only wipes the short-term conversation buffer. "reset to defaults" is a full factory reset, it wipes every command you've customised in the Settings GUI too, so use it deliberately.


Troubleshooting

Common issues & fixes

Most problems have a simple cause. Check these before anything else.

!

Friday doesn't hear me / no response to voice

Almost always a microphone configuration issue, not a bug.

🔧
1. Settings → System → Sound → Input, confirm your mic is selected as the default device.

2. Make sure the microphone volume is above 50%.

3. Check microphone permission: Settings → Privacy → Microphone → toggle on for desktop apps.

4. Confirm you said one of your configured wake words first, commands don't fire without one.
!

Friday crashes on startup

A crash on launch usually means a missing or corrupted file inside the install folder, or the installer didn't complete successfully, occasionally the case when a PyInstaller one-file build's temp extraction gets interrupted by another process (antivirus included).

Re-download Friday-Setup.exe from the GitHub Releases pageand the Downloads page and run it again. The original download may have been interrupted or corrupted mid-transfer.
!

Voice recognition is inaccurate

• Move closer to your microphone.
• Reduce background noise (TV, music, fans).
• Speak at a steady moderate pace, not too fast, not exaggerated.
• Switch to a headset microphone if using a built-in laptop mic.
• If a specific phrase keeps getting misheard, add your own wording for it in the Settings GUI instead of fighting the default one.
!

No sound / Friday speaks but I can't hear it

Check that speakers or headphones are connected and set as the default playback device in Windows Sound settings. Confirm system volume isn't muted. Friday plays audio through the Windows default output device.

!

Antivirus or Windows Defender flagging Friday

Friday's installer and its frozen executable are both unsigned, there's no commercial code-signing certificate. Some antivirus engines heuristically flag unsigned executables, especially ones built with PyInstaller or Nuitka, as potentially unwanted programs. Friday is not malware.

Add the Friday install folder as an exclusion in your antivirus software to prevent interference. You can verify authenticity by comparing the SHA-256 hash of your download against the hash published on the GitHub Releases page.
!

Friday can't find a newly installed app

The app catalogue is built when Friday starts and doesn't automatically update while running. After installing a new application, say "rescan" or "rebuild" to trigger a fresh catalogue scan.


FAQ

Frequently asked questions

The core loop does, Moonshine (speech recognition) and Piper (text-to-speech) both run entirely on your device. Optional features that reach out to a service, news, weather, email/Discord/Telegram checks, or AI chat via Duckduck Go/Groq, do need a connection, but nothing about hearing or responding to a command does.

Say "configuration" or Click on the friday Icon in the tray to open the Settings GUI, find the tab that matches the command (see the Customising section above), and edit the phrase list directly. Friday checks every new phrase against every other command list first, so you can't accidentally create a conflict.

Yes, wake words are just another editable list, on the GENERAL tab of the Settings GUI. You can add as many as you like; Friday responds to any one of them.

"rescan" (or "rebuild") only rebuilds the installed-app catalogue, say it after installing something new. "clear memory" wipes the short-term rolling conversation buffer, not anything permanent. "reset to defaults" is the big one, a full factory reset that also wipes every command phrase you've customised in the Settings GUI, so use it deliberately.

Yes. Just download and run the installer on each PC. There's no per-machine license, no activation, and no account needed. Each install builds up its own local config and TTS cache over time.

It grows slowly because it only stores audio for unique phrases Friday has actually spoken. In practice it stabilises quickly since daily usage repeats many phrases, greetings, acknowledgements, common responses. It's never cleared automatically, which is by design: Friday only gets faster over time.

No. Audio is processed locally by Moonshine and discarded after recognition, it's written to disk and never sent anywhere. Any credentials you configure (email app passwords, API keys) are stored DPAPI-encrypted on your own machine, readable only under your own Windows login. See the Privacy page for the full policy.

Download the new Friday-Setup.exe from the GitHub Releases page and run it, it updates the files in place. Because Friday tracks what shipped last time separately from what you've changed, every command or setting you've customised survives the update automatically; only genuinely new defaults get added.

Yes, entirely free. No paid version, no premium tier, no subscription, no ads. Friday is an independent personal project, distributed freely via GitHub.

The current build ships with an English Moonshine model. Other languages aren't officially supported in the current release.


Uninstalling

Removing Friday

Because Friday is installed through Inno Setup, removing her the right way is a normal Windows uninstall, not a manual folder deletion.

To completely remove Friday

Go to Settings → Apps → Installed apps (or Control Panel → Programs and Features) and uninstall Friday, or run unins000.exe from inside the install folder directly. This removes the application files and Start Menu shortcuts the installer created.

The uninstaller only removes what it installed. It does not remove a shortcut you manually placed in shell:startup, or the %APPDATA%\Friday folder Friday creates at runtime for her config and TTS cache, delete both yourself if you want a completely clean removal.
If you want to move Friday rather than remove her, move the entire install folder to the new location and update your shortcut paths (including the one in shell:startup) to point to the new Friday.exe. The TTS cache inside the folder moves with it automatically.

Ready to get started?

Download Friday from GitHub, free, offline at her core, and set up in under five minutes.