Wispr Flow
Wispr Flow is a voice-to-text AI that turns speech into polished, formatted writing inside any application on Mac, Windows, iOS and Android. It strips filler words, punctuates and formats automatically across 100+ languages.
What is Wispr Flow?
Wispr Flow is the dictation product of Wispr AI, Inc., a San Francisco company that calls itself “The Voice Interface Company”. You hold a key, you speak, and formatted text lands wherever your cursor already is: Gmail, Slack, iMessage, Notion, a terminal, a code editor, or an AI tool such as ChatGPT, Claude or Cursor. There is no plugin to install and no per-application setup.
The pitch is intent rather than raw transcription. Built-in dictation writes down every hesitation, false start and mid-sentence correction and leaves you to clean it up; Flow removes them, punctuates, and formats lists, paragraphs and emails as you talk. Say “let's meet at 5… actually 6pm” and you get “Let's meet at 6pm.” The company claims 220 words per minute against 45 on a keyboard, and a 90% zero-edit rate, its own benchmark placing OpenAI at 71%, ElevenLabs at 63% and Siri at 52%.
Flow runs natively on macOS, Windows, iOS and Android, with a web admin portal for teams. It supports over 100 languages, including regional variants such as British, Canadian and Swiss orthography, Simplified and Traditional Chinese, Cantonese, and code-switched Hinglish. Personalisation is central: a dictionary that learns your names and jargon, voice-triggered snippets, and per-application tone so a work chat does not read like a legal brief.
A second product, Wispr Flow Notetaker, adds meeting notes with speaker identification, enhanced summaries, action items and cross-meeting question answering. It handles Google Meet, Teams, Huddles and in-person conversations, and notably without a bot joining the call, since everything runs on the device. Notetaker is Mac-only for now.
Privacy is positioned as a differentiator: model training is opt-in and reversible, Zero Data Retention is available, and Wispr holds zero-retention agreements with every third-party AI provider. Advertised certifications are SOC 2 Type II, ISO 27001 and HIPAA, with a Business Associate Agreement offered on all plans. All data is processed and stored in the United States. Flow is free for 2,000 words a week; Pro starts at 12.00 USD per user and per month billed annually.
What it does
- Dictate polished, ready-to-send text into any application, with no plugin or integration
- Strip filler words, repetitions and mid-sentence corrections automatically
- Format numbered lists, paragraphs and structured emails while you are still speaking
- Dictate in 100+ languages with per-session detection and regional orthography
- Capture meetings with speaker identification, summaries and action items
- Teach Flow your names, jargon and reusable voice-triggered snippets
- Push meeting notes into Claude, ChatGPT and Cursor over MCP
When to use Wispr Flow / When not to
A quick filter to help you decide if Wispr Flow is the right fit.
When to use Wispr Flow
- Professionals who write all day across email, chat, documents and code editors
- Software engineers dictating inside an IDE, with camelCase and snake_case handling and file tagging in Cursor and Windsurf
- Customer support and sales teams facing high message volume, where Clay reports 20% more customer calls per day
- People for whom typing is painful or impossible, including users living with Parkinson's, arthritis or age-related mobility loss
- Multilingual speakers who switch languages, need regional spellings or work in code-switched Hinglish
When not to use Wispr Flow
- Teams that need meeting notes outside macOS: the Notetaker is Mac-only today, with Windows announced and mobile merely on the roadmap
- Developers who want to build on a public API, since Wispr has sunsetted its own and routes export through the admin portal instead
- Organisations requiring on-premise deployment or offline, on-device transcription, because every transcription runs in the cloud
- Buyers bound by European data-residency rules, as all customer data is processed and stored in the United States
- Smaller teams expecting SOC 2 Type II or ISO 27001 coverage without signing an Enterprise contract
How to use Wispr Flow
A typical end-to-end flow, from setup to results.
- Download the application for Mac, Windows, iPhone or Android from the Get started page
- Sign in with an email address or an existing Google, Apple or Microsoft account, or through SSO, with no credit card required
- Select the languages you actually speak in Settings, since picking a single one beats auto-detect for accuracy
- Hold the dictation key and speak naturally, hesitations and self-corrections included
- Let Flow clean, punctuate and format the text before it lands in the active application
- Add names and jargon to your personal dictionary, or let Auto-add capture the corrections you type
- Create voice-triggered snippets for the emails, links, addresses and bios you retype constantly
- Set your data preferences in Settings > Data and Privacy: model training, Cloud Sync, HIPAA BAA and transcript auto-deletion
- On Mac, start the Notetaker with one button to capture meetings, including calls absent from your calendar
- For a team, invite colleagues from the admin portal and share the dictionary and snippets
Pros & Cons
Pros
- Works in every text field on every device, with no plugin, integration or per-application setup
- Output arrives already formatted and ready to send, with a claimed 90% zero-edit rate
- A genuinely permanent free plan rather than a trial, with no countdown and no credit card
- Unusually granular privacy controls up to Zero Data Retention, backed by zero-retention agreements with every third-party AI provider
- HIPAA Business Associate Agreement reachable from an individual account, not just Enterprise
- Over 100 languages with regional variants and support for code-switching
- 50% off for students and educators, and reduced pricing for non-profit organisations
Cons
- The Notetaker is Mac-only, with Windows announced and iPhone and Android merely on the roadmap
- No public API whatsoever: Wispr has sunsetted it, and export now runs through the admin portal
- No on-premise option and no European hosting region, since every byte is stored in the United States
- Transcription always happens in the cloud and never locally, even with Privacy Mode enabled
- SOC 2 Type II and ISO 27001 stop at the Enterprise plan, and the subprocessor list is available only under NDA
- Free-plan quotas are inconsistent: 2,000 words a week on desktop, 1,000 on iPhone, unlimited on Android
- Android exposes only a subset of the Data & Privacy controls, with no transcript auto-deletion
Pricing & Plans
Wispr Flow provides a permanent free plan rather than a time-limited trial: it requires no credit card and covers 2,000 dictated words per week on desktop, 1,000 on iPhone and unlimited on Android, with the Notetaker included on Mac. The lowest paid entry point is Flow Pro at 12.00 USD per user and per month when billed annually, rising to 15.00 USD per user and per month on monthly billing. The Enterprise plan is quoted on request and carries volume discounts.
- dictation in 100+ languages on Mac
- Windows
- iOS and Android
- Notetaker on Mac with speaker identification
- MCP access
- calendar and Slack connection
- HIPAA-ready handling and model-training opt-out
- capped at 2
- unlimited dictation
- more meetings kept for longer
- access to advanced AI thinking models
- early access to new features
- prioritised support tickets
- centralised billing and user management
- shared dictionary and snippets
- enterprise-grade security and admin controls
- SOC 2 Type II and ISO 27001 compliance
- enforced HIPAA-ready BAA
- complimentary seats for IT admins
- audit logs
- MDM and domain management
- SAML SSO and SCIM
- dedicated support
Data, GDPR & hosting
A consolidated view of how Wispr Flow handles your data.
GDPR overview
GDPR is addressed concretely rather than by slogan. Section 10 of the privacy policy enumerates the rights of EU and EEA residents, namely access, rectification, erasure, objection, restriction, portability and withdrawal of consent, names support+privacy@wisprflow.ai as the point of contact, and restates the right to complain to a supervisory authority. A Data Processing Addendum is available: it incorporates the June 2021 EU Standard Contractual Clauses, adds a UK Addendum for the UK GDPR, and carries technical and organisational measures in Annex 3. GDPR data export is supported. Two limits deserve attention: no Article 27 EU representative is named anywhere on the site, and Wispr operates no European processing location, so all customer data is handled in the United States wherever the user sits. The privacy policy is dated 19 August 2026.
Who owns the data?
Wispr's Terms of Service are unusually blunt: all your content remains yours, and Wispr does not own it. You keep every ownership right in the audio and text you submit, and you own the generated output outright, Wispr formally assigning to you whatever rights it might hold in it. In exchange, you grant Wispr and the third-party services acting on its behalf a worldwide, non-exclusive, royalty-free, sublicensable licence to store, reproduce, modify, transmit and display that content, limited to providing, improving and protecting the service. The Data Controls page states the commercial counterpart plainly: Wispr never sells your data, because its business is selling software instead.
Reuse rights
Since you retain ownership of both what you dictate and what Flow produces, you may reuse either freely, commercially or otherwise, without asking anyone's permission. What is bounded is Wispr's own reuse. Training on your content is a choice you make in Settings > Data and Privacy, a toggle formerly called Privacy Mode; switching it off alongside Cloud Sync yields Zero Data Retention, where nothing is trained on and nothing is stored server-side. The third-party model providers named in the terms, OpenAI and Anthropic, are contractually barred from training on your data and delete it after 30 days. Google Calendar, Gmail and Workspace data are never fed into training nor used for advertising, and Wispr collects no voiceprint or other biometric identifier. Usage statistics such as your dictated word count are gathered regardless of these settings.
Data retention & training
Hosting summary
Wispr Flow is multi-tenant SaaS hosted entirely with a major US cloud provider that the company does not name. Wispr operates no data centres of its own; the provider's facilities are independently attested to SOC 2, ISO 27001 and PCI DSS, and no production customer data sits on employee devices. All customer data is processed and stored in the United States regardless of where the user is located, and Wispr operates no European or other regional processing location, so there is no cross-border storage; EU and UK transfers rely instead on the June 2021 Standard Contractual Clauses in the Data Processing Addendum, with a UK Addendum available. There is no on-premise version. Data is encrypted in transit over TLS and at rest on Wispr's servers, and transcription always happens in the cloud rather than on the device. Notetaker transcripts are held in Wispr's private cloud for cross-device syncing, while encrypted meeting audio may be kept briefly on the device or in cloud storage before automatic deletion.
Things to keep in mind
Risks and trade-offs to weigh before adopting Wispr Flow.
- Everything you dictate travels to a US cloud before it comes back as text: transcription is never local, even with Privacy Mode enabled
- The Notetaker records conversations, and Wispr states explicitly that it does not determine whether notice or consent is legally required, leaving that liability with you
- Ambient capture is easy to forget: Context Awareness reads text on your screen, and a meeting recorder that needs no bot leaves other participants no visible cue
- Deleting your account does not delete everything, since aggregated statistics, credits, referrals and a compliance audit record are retained by design
- Leaning on automatic cleanup can quietly erode the habit of composing and editing your own sentences, because the tool decides what you meant and is not always right
- Dictating sensitive material in an open-plan office or a public place exposes it to everyone within earshot, a risk no encryption setting addresses
- Marketing and documentation do not fully agree on certification: the site advertises SOC 2 Type II while the help centre evidences a SOC 2 Type I report and Stage 1 ISO 27001 documentation
Setup & Integrations
Technical difficulty
Very low for an individual. You install a native application, sign in with an email address or an existing Google, Apple or Microsoft account, and start dictating; nothing is added to the applications you write in, and Wispr claims teams are running within minutes. The only optional settings are your active languages, your dictionary and your data preferences. Effort shifts to IT only at Enterprise level, where SAML SSO, SCIM, MDM and domain management require a proper identity and device-management setup.
Deployment
Integrations
Supported languages
Behind Wispr Flow
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Alternatives
Tools that compete with or complement Wispr Flow.
Frequently asked questions
Do I need to install anything in Gmail, Slack or my code editor?
How is Flow different from Siri or built-in voice typing?
Is Wispr Flow really free?
Which platforms does it run on?
Is my voice data used to train Wispr's models?
Where is my data hosted?
Does Wispr Flow expose an API?
Are there any discounts?
Is there a minimum age to use Flow?
Does the Notetaker send a bot into my meetings?
Should you pick Wispr Flow?
Wispr Flow is a mature product rather than a demo: 81 million USD raised according to the company's own announcement, native applications on four platforms, independent certifications, and a support operation claiming more than 100,000 resolved tickets. Its central bet pays off. Because Flow types wherever the cursor is, it replaces the half-dozen voice modes scattered across a phone, an IDE and a chat client with a single consistent behaviour, and what it produces is genuinely ready to send rather than raw transcription waiting to be cleaned up.
The privacy posture is unusually detailed for a consumer tool. Model training is opt-in, Zero Data Retention is a real setting, no voiceprint or biometric identifier is collected, and a HIPAA Business Associate Agreement is reachable even from an individual account. The fine print still matters: transcription always runs in the cloud, every byte is stored in the United States with no European region on offer, the subprocessor list is available only under NDA, and SOC 2 Type II and ISO 27001 stop at the Enterprise plan, while the help centre only evidences a SOC 2 Type I report and Stage 1 ISO documentation.
The remaining gaps are mostly maturity gaps. The Notetaker, the most interesting recent addition, is Mac-only. There is no public API at all, which rules Flow out of any automated pipeline. Free-plan quotas differ oddly between desktop, iPhone and Android.
For anyone who writes for a living, the free plan is generous enough to answer the only question that matters, whether dictation actually fits the way you work, before spending anything. If it does, 12.00 USD per user and per month on annual billing is a modest step up. Buyers with European data-residency obligations, or a need for an API, should look elsewhere.
- Choosing a selection results in a full page refresh.
- Opens in a new window.