Vehla started as a personal tool. Then friends started asking for the build. Then strangers. That is the whole origin story, and it explains most of the decisions in the app better than any positioning document would.
How this started
We were not trying to build a launcher. We were trying to stop leaving the app we were already in. The best writing tool on a Mac is whatever window your cursor is currently in — Mail, Notes, an IDE, a half-finished Slack message. The worst place to write is a chatbot in a browser tab, where you tab away, lose the thread, paste, edit, paste back, and forget what you were doing.
So the first version was small: a hotkey, a palette, and a handful of text actions that operated on the selection and put the result back where it came from. Everything since has been an argument about what else deserves to be behind that same hotkey.
The tab-switching tax
Look at a working Mac and count the menu-bar icons. A window manager. A clipboard manager. A file shelf. A menu-bar cleaner. A screenshot-and-OCR tool. A color picker. A dictation app. A text expander. An uninstaller. A notes app. Maybe a notch utility. Each one is good. Each one was a reasonable purchase. Together they are a dozen preference panes, a dozen update mechanisms, a dozen sets of keyboard shortcuts that collide with each other, and a dozen small monthly charges you have stopped noticing.
We counted fourteen categories on our own machines. That list is now on the site, with the specific apps named, because the honest version of our pitch is not "we are better than Magnet." It is "you probably do not want to run fourteen of these."
How bundles fail
Consolidation is the obvious move, and it usually produces something worse than what it replaced. Three failure modes we watched for.
Fourteen mediocre features. Each surface gets built to eighty percent, which is exactly the level at which you keep the original app installed anyway. Now you have fifteen apps.
A settings screen nobody can navigate. Fourteen apps' worth of preferences in one window, organized by the engineering team's mental model rather than yours.
A palette that gets slower every release. The core loop is the product. Every feature that adds a millisecond to opening the palette is taking from the thing that made you install it.
We have not fully escaped the second one. The settings window is large, and it is the part of the app we are least happy with.
Our three rules
Two. Anything you do not use can be hidden. The palette is a personal object: hide actions, favorite the ones you run, rename them with aliases. A feature you have never opened should cost you nothing but disk.
Three. The defaults are the work. Not the settings, the defaults. The exact tone of "Rewrite — Confident", the right level of brevity for "Summarize", the shape of the event that "Generate Calendar Event" produces. Anyone can ship a prompt box. Tuning 73 actions so you do not have to think is the actual labour.
Local-first as a default
You should not have to "go private" to feel safe. That is why local mode is not a toggle buried in a privacy pane — it is a route in the palette footer, next to the cloud providers, named in every prompt.
The concrete version of this belief is a decision that costs us complaints: if you have selected Local AI and the model is unavailable, we show an error instead of falling back to a cloud provider. People do occasionally write in about this. It stays, because silent fallback would make the entire feature a lie.
Buy, don't rent
Vehla Free is free forever, no card, and it is a usable app rather than a countdown. Vehla Pro is $29.99 once, two Macs, updates for life. There is a 7-day Premium trial that also does not ask for a card.
This is a worse business model than a subscription. It is also the model that matches what the app is: a utility you install and forget about, like a window manager. A tool you use every day should not require a renewal decision every month. If the app is worth $29.99 to you, pay once and stop thinking about us.
What we say no to
Most feature requests we receive are good ideas. We ship the ones that make Vehla feel sharper rather than heavier, which means saying no to a lot of reasonable requests from people who were kind enough to send them. The recurring ones we have declined: an iOS companion, a team tier, a hosted model service, and a plugin system with fewer permission checks than the current one.
Some of those we may build eventually. The hosted model service we will not — the moment we are in the middle of your prompts, every claim on the privacy page needs an asterisk.
What's next
The roadmap is short on purpose: keep the palette fast, make local and speech models easier to manage, improve the action and extension system, and polish the parts of macOS integration that make the app feel native rather than cross-platform. Nothing on that list is a new category.
We are a tiny team in Ottawa and every email gets a reply, usually within a day, often from whoever wrote the code in question. If you have an opinion about any of the above, send it.