Wispr Flow

Wispr Flow is system-wide AI dictation: hold a hotkey, speak, and the text lands in whatever app has focus — already punctuated, with the ums and false starts removed. It is the only tool in this category with first-party apps on all four of macOS, Windows, iPhone and Android. Every word is transcribed in the cloud; there is no on-device mode at any price.

What You Get#

• Dictation into any app: Works at the OS level, so Scrivener, Word, Google Docs, Slack and email all behave the same. No extension, no separate notepad, no paste step.
• AI cleanup and tone matching: An LLM pass rewrites raw speech into finished sentences and adapts register to the app you are in — a chat reply reads differently from a manuscript paragraph.
• Command Mode: Edit by voice. Select text and say what to change instead of re-dictating the passage.
• Custom dictionary and snippets: Teach it character names, invented place names and recurring phrases so they stop coming back mangled.
• 100+ languages for dictation input, plus context awareness that reads on-screen text to improve accuracy.

Who It's For#

• Authors who draft on more than one device: Dictating a scene on a phone walk and finishing it in Scrivener is the workflow this covers better than anything else here.
• Authors who want zero cleanup: If you have tried raw speech-to-text and hated editing the transcript, the LLM pass is the actual product.

Pricing: Free tier caps vary by platform — 2,000 words/week on desktop, 1,000/week on iPhone, unlimited on Android. Pro is $12/month billed annually or $15/month billed monthly. Enterprise is quote-only and adds SSO, SCIM and audit logs.

Strengths#

• Platform breadth: Mac, Windows, iPhone and Android, all native. Competitors cover one or two.
• Cleanup quality is the differentiator: What lands on the page needs materially less editing than raw speech-to-text — this is what you are paying for.
• Well funded: A $280M Series B at a $2B valuation in August 2026 makes vendor abandonment unlikely. An in-house speech model, Canto, was previewed at the same time but has not shipped.

Limitations#

• Cloud-only, permanently: Wispr states transcription always happens in the cloud. Authors under NDA or handling client manuscripts have no offline option.
• Context Awareness captures screenshots: On by default on Mac and Windows, it sends app contents, on-screen text and screenshots with each dictation. It can be turned off.
• Compliance history: Wispr invalidated its own SOC 2 Type II and ISO 27001 certificates in March 2026 after the auditors were implicated in a mass-produced-reports scandal, then re-audited with A-LIGN. A clean SOC 2 Type I landed in April 2026; Type II has still not been issued. A HIPAA BAA exists but only on request through sales, not on Pro.
• Ratings are split: 2.7/5 from 47 Trustpilot reviews against 4.8/5 from 8,500+ iOS App Store ratings. Trustpilot complaints centre on accuracy dropping after the trial ends, unhonoured referral rewards and slow support.
• Platform gotchas: No iPad, no Linux, no web app, and Windows is x64 only — ARM laptops are unsupported.

Alternatives#

Superwhisper is the direct trade: cheaper, and it offers genuine on-device transcription, but no Android app. MacWhisper is a one-time purchase with offline dictation if you only work on a Mac. Voice In is far cheaper but browser-tabs only. Choose Wispr Flow for reach and cleanup; choose the others if your manuscript cannot leave your machine.