Breaking: Willow raises $4.2M to build voice-first interfaces Speak naturally, Willow writes it ~3x more accurate than built-in dictation 200ms speech-to-text latency Early users: Uber · Heidi Health · Zego Backed by Y Combinator + BoxGroup Now on Mac · Windows · iPhone Breaking: Willow raises $4.2M to build voice-first interfaces Speak naturally, Willow writes it ~3x more accurate than built-in dictation 200ms speech-to-text latency Early users: Uber · Heidi Health · Zego Backed by Y Combinator + BoxGroup Now on Mac · Windows · iPhone
Company Dossier · Voice-First Computing

Willow.

The voice interface quietly trying to retire your keyboard. You talk. It writes. Across every app, on Mac, Windows and iPhone.

San Francisco Founded 2024 Y Combinator Seed · $4.2M ~29 people
Willow voice dictation product
WILLOW, ON THE JOB. A cursor blinks in an empty document. Someone starts talking. The words arrive already punctuated - filler stripped, tone matched. The keyboard just stands there.
The Scene

The keyboard has had a very good run

Since 1868 we've all agreed to press small squares in a deliberately awkward order. Willow thinks the agreement is up for renegotiation.

Open a laptop in any coffee shop in San Francisco and you'll hear the same thing: a soft clatter of keys, the sound of people translating thoughts into finger movements. Now look closer. A few of them aren't typing. They're talking - quietly, unhurried - and full sentences are landing in their email, their Slack, their code editor, already clean. That's Willow at work.

Willow is a voice dictation tool that turns speech into finished text across any application. Not the robotic dictation you remember from a decade ago, where you had to say "comma" and "new paragraph" and then spend ten minutes fixing the wreckage. Willow removes the "ums," adds the punctuation, matches how you actually write, and hands you a paragraph you'd be happy to send. The company's own line for it is refreshingly plain: Speak naturally, Willow writes it.

A lot of communication takes place outside meetings.
Allan Guo, co-founder & CEO, on why Willow chose dictation over the crowded AI-notetaker market

That sentence is the whole thesis. The AI world spent two years building tools to summarize meetings. Willow noticed that the real firehose of writing - the emails, the messages, the documents, the prompts fed to other AIs - happens everywhere else, all day, one keystroke at a time. If you could speak all of it at the speed of thought, you'd get a chunk of your week back.


By The Numbers

The pitch, in figures

$4.2M
Seed raised
~200ms
Speech to text
~50%
MoM user growth
100+
Languages

Willow markets a roughly 3x accuracy jump over built-in dictation and claims users write about 4x faster - reclaiming something like five hours a week. Numbers like these are self-reported, so read them as ambition backed by traction rather than audited fact. The growth curve, at least, is the kind that makes investors return calls.


What You Can Actually Do

Talk. Edit. Ship.

Willow isn't one app so much as a layer that sits on top of everything you already use.

Desktop · Mac & Windows

Dictate anywhere

Works across every app with smart formatting, filler-word removal, and a dictionary that learns your names and jargon. Offline mode and a "Hey Willow" assistant included.

iOS Keyboard

Type by voice - then fix it

A voice keyboard for every iOS app, with a full keyboard right there for edits. You can correct what you said instead of re-recording the whole thing.

AI Mode

Mumble in, polish out

Speak a rough, half-formed note and Willow reshapes it into a clean, context-appropriate message for email, chat, or docs.

For Developers

Voice in the IDE

Tuned for coding environments like Cursor and AI IDEs - it handles technical vocabulary and hands-free prompting without turning "async" into "a sink."

Whisper Mode

Dictate quietly

Speak under your breath in a shared office or cafe and Willow still keeps up - built for the awkward reality of talking to your computer in public.

For Teams

Roll it out

A team and enterprise tier for organizations adopting voice dictation at scale, with the security posture to match.


The Founders

Two dropouts, ten-plus pivots

Willow is the thing that finally worked. Before it, there was a healthcare startup, a lot of dead ends, and a year in the trenches.

Co-founder & CEO

Allan Guo

Left Stanford as a freshman, watched his first startup fail, and pivoted more than ten times before landing on voice. Lists his own title as "Chief Keyboard Killer." He's 19.

Co-founder & CTO

Lawrence Liu

Guo's co-founder and the other half of the Stanford exit. Together they took Willow through Y Combinator and turned a decade-old idea - dictation - into something people actually keep open.

We pivoted over 10 times, spent a year in the trenches, and learned how to build.
— Allan Guo, announcing Willow's $4.2M round

The Business

Consumer habit, enterprise wallet

Mission

Make communication faster and more natural by letting people speak instead of type - turning voice into polished text across any app, on any device. The longer bet: a voice-first interface where speaking is how you write and steer your computer.

How it makes money

  • MODEL Freemium SaaS subscription
  • CONSUMER Paid plan (~$144/yr reported)
  • TEAMS Enterprise tier for organizations
  • REACH B2C and B2B, same product core

Who's using it

  • PEOPLE Founders, writers, developers, clinicians
  • ENTERPRISE Uber · Heidi Health · Zego
  • SECURITY SOC 2 Type II · HIPAA-compliant
  • TECH Llama models for formatting & personalization

Who's in the ring

The obvious rival is Wispr Flow - same rough price, different fit. Then Superwhisper, Aqua Voice, MacWhisper, and the free built-in dictation from Apple, Google and Microsoft. Willow's wager is that "good enough and free" loses to "fast, accurate, and editable."


The Cadence

A year of shipping

MAY 2025
Closes a $4.2M seed led by BoxGroup, with Y Combinator, Burst Capital, and angels including HubSpot's Dharmesh Shah, Reddit's Alexis Ohanian, and Instacart's Max Mullen.
NOV 2025
Launches the iOS voice keyboard - type across every app, and actually edit what you said.
JAN 2026
Ships Willow for Windows.
FEB 2026
Releases Willow for Developers, aimed at Cursor and AI IDEs.
MAR 2026
Launches Willow for Teams for organizations.

Watch & Listen

See it talk


The Scene, Revisited

Back in the coffee shop

Same cafe. Same clatter of keys. But now the person who used to hunt-and-peck through email is leaning back, coffee in hand, talking in a low voice while the messages write themselves - filler stripped, tone matched, sent. The keyboard is still on the table. It's just not doing much anymore.

Willow hasn't killed the keyboard - not yet, and maybe not ever. But it has done something quieter and more interesting: it made talking to your computer feel normal. Two Stanford dropouts, ten-plus pivots, and a $4.2M bet later, the sound of work is starting to change. Less clatter. More conversation.

Speak naturally, Willow writes it.
— Willow's promise, and increasingly its habit

Share this dossier

Send Willow to someone still typing the hard way.


Find Willow

Links & sources