Screen & audio recording
Record the full screen, a single window or audio only, together with the microphone and system audio. Pause and resume as you go, and stop from anywhere with a global shortcut.
Recording transcripts and AI summaries on Windows
Hours of recordings. Minutes of decisions.
ScreenBrief records your meetings and screen, then turns them into a transcript, a summary, action items and a chat. Check answers against citations that take you to the relevant minute of the recording. Your library is stored on your disk. Choose local models or your own provider APIs for processing. The first 5 transcriptions are free; Pro costs 299 PLN once or 29 PLN a month.
Once or monthly299 PLN once or 29 PLN a month
Data on your disklocal database, no vendor cloud
Timestamped citationsone click takes you to the moment
Windows 10/11available in Microsoft Store
See ScreenBrief in action
From a recording to clear takeaways in a minute
See the transcript, summary and answers that point to the exact minute of your recording.
Music and sound effects: original, made for ScreenBrief. Voice: ElevenLabs.
Try it free in Microsoft StoreThe app in action
Scroll to see how ScreenBrief works
- 01
Dashboard
- 02
Library
- 03
Recording
- 04
Ask AI
- 05
Search
You see the whole week on one screen: the key numbers, recordings being processed, open action items and the latest decisions.
You keep recordings in order with filters (calls, screens, processing, starred), tags, participants and statuses, with no manual housekeeping.
Next to the player you get a speaker-labelled transcript, short and detailed summaries, action items with due dates and the decision log.
The chat answers only from the recording and backs every claim with a quote pointing to the minute.
You search all recordings at once: utterances, screen events and decisions. Every result takes you to the exact second.
Features
Capture. Understand. Act.
File & link import
Drop recordings from Zoom, Teams or a voice recorder (mp4, mov, mkv, webm, mp3, wav and more), add them without copying, or paste a link to a file.
Transcription: local or via API
Transcribe locally with Whisper (on the CPU or GPU, with three model sizes) or through an external provider: OpenAI, Groq, Deepgram, AssemblyAI.
Speaker recognition
ScreenBrief splits the conversation into speakers on its own and assigns them the real names spoken in the call. Names and the transcript can be corrected by hand.
A brief after every meeting
You get short and detailed summaries, tasks with an owner and a due date, and decisions stamped with the minute they were made. Regenerate any of it with your own prompt.
Auto-tagged timeline
The timeline splits a recording into lanes: speakers, screen activity, pricing threads, objections and decisions. No more scrubbing blind.
Screen analysis (AI Vision)
Turn on frame analysis and the text visible on screen becomes events and lands in search.
Chat with quotes
Ask about one recording or a whole folder, such as every meeting with one client. Every answer arrives with a quote and the minute.
Full-text search
One search box covers utterances, screen events and decisions across all recordings. A result takes you to the exact second.
Export & copy
Export a recording to Markdown, SRT, VTT and PDF, and action items to your calendar (ICS). Copy the transcript, summary and chat answers in one click.
AI cost control
A monthly dollar cap is checked before every AI call. You see costs split across transcription, summaries and chat, and can assign a different model to each task.
Languages
The interface runs in Polish and English. Transcription handles about 100 languages with auto-detection, and the AI output language is set separately.
Ask AI
Every answer is backed by the recording.
- It answers only from the selected recording, not from the model's general knowledge.
- Every claim carries a quote with the minute; one click seeks the player to that moment.
- When the recording doesn't support a confident answer, you get a clear notice instead of a guess.
- You start with ready-made questions: about pricing, objections and what was shown on screen.
Search
Every word has an address.
- You search the full text of every recording: utterances, events and decisions.
- The phrase is highlighted in the snippet: you see context before you click.
- A result links to the exact second in the recording.
- “We talked about this in March” stops being a problem.
Privacy
Your recordings stay on your computer.
ScreenBrief is a local app, not a cloud service. No accounts, no syncing, no “analytics”. Only what you choose to send reaches the AI models, through your own keys.
Local databaseThe SQLite database and recording files live on your disk, in your user directory.
Local transcriptionTranscribe locally (Whisper on your hardware) or with an external provider via API. The choice is yours.
Your AI keysSummaries and chat use your own API keys or local Ollama, capped by a spend limit you set.
Contact-data redactionOn by default: email addresses and phone numbers are stripped from the transcript before any text reaches a model.
How it works in practice: transcription where the recording never leaves your computer →
Guides
Start with the problem you need to solve.
Focused answers about local transcription, AI meeting notes, and screen recordings on Windows.
01
Offline meeting transcription
What really stays on your computer and when data can reach an external model.
Read the guide → 02Local AI meeting notes on Windows
A workflow without a bot joining the call.
Read the guide → 03Transcribe a screen recording
Connect spoken words, screen context, summaries, and the exact point on the timeline.
Read the guide → 04Summarize a recording with Ollama
A summary and action items from a model running on your own computer, with no API key.
Read the guide → 05Teams, Zoom and Meet recordings
Where the recording file lives and how to load it for transcription without a bot in the call.
Read the guide → 06Whisper without a terminal
Transcription on your own computer without Python or the command line: which model to pick and how big it is.
Read the guide → 07Find a decision in a recording
Library-wide search, chat that cites the exact minute, and one question across a whole folder.
Read the guide → 08Transcribe a lecture
A phone recording into a transcript, notes and subtitles; up to four hours per file.
Read the guide → 09Job interviews
A speaker-labelled transcript, the candidate's consent, and deletion once the role is filled.
Read the guide → 10SRT subtitles
A timestamped transcript as an .srt or .vtt file, with your corrections in it.
Read the guide → 11Local model benchmark
Three Ollama models, one meeting with traps, three runs each: the withdrawn task survived 8 of 9.
Read the guide →FAQ
Frequently asked questions
Is this a subscription?
Your choice. Pro is 299 PLN once, or 29 PLN a month that renews until you cancel it. The features are the same, and you stop a subscription in the Stripe customer portal. The choice lives in the app from version 0.8.0.
Can I try it before buying?
Yes. The first 5 transcriptions are free. Pick local Whisper and they need no API key at all. Once that allowance is used up your existing recordings and transcripts stay available, and buying Pro in the app unlocks new transcriptions.
How many computers does one license cover?
One. A Pro license is signed for a specific installation, so a second computer needs its own. After a reinstall on the same computer it comes back by itself. The license renews over the internet roughly once a month, and between renewals ScreenBrief works offline.
Where is my data stored?
Locally: a SQLite database and recording files in your user directory. ScreenBrief never uploads recordings or transcripts to any vendor server. If you point it at an external AI provider yourself, that provider receives (under your own key) the audio for API transcription, or excerpts of the transcript for summaries and chat. With local Whisper and local Ollama, nothing leaves your computer.
The longer answer: what stays on disk, and what still leaves →
Do I need API keys?
Not for transcription: local Whisper runs without any key. Summaries and chat do need an AI provider, so either your own key (Anthropic or OpenAI, for example) or local Ollama, which needs no key either. Spending is guarded by a hard cap you set yourself.
Do I need a powerful computer?
No. Local transcription runs on the CPU alone, and if you have a graphics card ScreenBrief can use it through built-in Vulkan or optional CUDA for NVIDIA cards. On weaker hardware you pick a smaller model, and you can move transcription entirely to an external provider via API so your machine does none of the work.
Which languages are supported?
Interface: Polish and English. Transcription: about 100 languages with auto-detection. The summary and AI answer language is set separately, so you can record in Polish and get the brief in English.
Which platforms does it run on?
ScreenBrief is currently available for Windows 10 and Windows 11 (x64) in Microsoft Store. You can purchase Pro later inside the app, either once or as a monthly subscription.
Pricing
Start free. Buy Pro once, or pay monthly.
The app is a free download and comes with 5 transcriptions. You unlock Pro from inside the app: 299 PLN once, or 29 PLN a month.
- one computer, no account, no sign-in
- data on your disk
- PL / EN interface
- 5 free transcriptions to start
Windows
Microsoft Storefree download
Free start
5 free transcriptions
Already subscribed?
Manage your subscription
The portal shows the state and the dates, changes the payment method and stops the renewal. Enter the email you bought with; Stripe sends the sign-in link there.
Collect decisions, not recordings.
ScreenBrief · Artflow Management