Willow gets the core idea right
Willow is a voice dictation app that lets you talk instead of type, and the push-to-talk flow is the right one: hold a key, speak, get text. If you’ve used it and liked the feel, you and I want the same thing. I built Invoke because I wanted that feel without two things Willow asks of you.
The first is a subscription. The second is the cloud.
The catch: cloud, and a monthly bill
Willow sends your audio off your machine to be transcribed, and it charges you every month to keep doing it. For a lot of people that’s a fine trade — it’s how most dictation apps work. But if you’re a developer dictating prompts and code into AI tools all day, two things start to bug you:
- Your words — which often include code, file paths, and internal project details — leave your machine.
- You’re renting a feature you use constantly, forever.
Invoke flips both. Transcription runs locally on your own GPU, and you pay once.
| Invoke | Willow | |
|---|---|---|
| Price | $49 one-time | Monthly subscription |
| Transcription | Local (on your machine) | Cloud |
| Your audio | Never leaves your device | Sent to their servers |
| GPU acceleration | NVIDIA CUDA, sub-second | N/A (cloud) |
| Push-to-talk | Yes | Yes |
| Works offline | Yes | No |
| AI reformatter | Yes, with project context | No |
What changes when transcription is local
When the model runs on your hardware, two things get better. It’s private — your audio is transcribed in memory on your machine and never uploaded. And on an NVIDIA card it’s fast, because there’s no network round-trip. You hold the key, speak, and the text is there before you’ve let go.
The flip side is honest: you need a decent machine. Willow’s cloud does the heavy lifting so a low-end laptop works fine. Invoke leans on your GPU. If you’ve got an NVIDIA card, that’s a feature. If you don’t, it falls back to CPU and runs a little slower.
Invoke is $49 once. Free 7-day trial, all features, no credit card.
The pricing math
A monthly dictation subscription is the kind of thing you forget you’re paying for. At roughly $10–12/month, you’re past Invoke’s one-time price inside half a year, and you keep paying after that.
Invoke is $49 once with a year of updates, or $79 for lifetime updates. After that there’s nothing to cancel. For a tool you use every single day, owning it beats renting it.
The part Willow doesn’t do
Invoke isn’t just transcription — it can reformat what you said into an actual prompt. It reads your project’s CLAUDE.md, package.json, and git history, so when you mumble “add a post endpoint for user off” it knows you meant authentication. That’s the difference between dictation and a tool built for developers. More on that in how I dictate code all day.
When Willow is the better call
If you’re on a thin laptop with no GPU, or you bounce between machines and want something that just works in the cloud, Willow’s model fits you better. No GPU requirement, nothing to install locally.
If you want your audio to stay on your machine, you have an NVIDIA GPU, and you’d rather pay once — that’s Invoke. See also voice to text with no subscription and the Wispr Flow alternative writeup, which covers the same cloud-vs-local tradeoff.
Try it
7-day free trial, no credit card. Download here.