What a €69 app has to refuse

A price you charge once is usually read as generosity. It is not. It is a constraint, and the constraint has teeth.

Someone who paid last year pays nothing this year and still receives this year's work. Everything I add has to be maintained forever on money from strangers who have not bought yet. That arithmetic is written out in The arithmetic of selling software once. This piece is the other half of it. The roadmap has to be shorter than a subscription's, which means there is a list of things I will not build.

I would rather publish the list than have you discover it. A refusal is also the only kind of claim you can check. "Privacy first" costs nothing to print. Declining a feature that customers ask for costs something, every time somebody asks.

No video

yapyap is audio only. No video capture, no video playback.

Saying yes would mean encode and decode paths on three desktop platforms, thumbnails, seeking, a player, and files an order of magnitude larger sitting in a library designed around audio. Every speech model I integrate afterwards would have to be tested against a container format that has nothing to do with speech. It would roughly double the surface I maintain in exchange for a thing screen recorders already do well.

Audio is the part I can do properly for one price, forever. Video is the part where I would quietly need your money again next year.

No team administration

No multi-seat admin, no organisation hierarchy, no team billing.

This is the refusal with the most money on the other side of it. Seats are how tools in this category get expensive: you sell to one person, then bill the company per head. To do that honestly I would need roles, provisioning, an audit trail, a billing back end, and an account system to hang all of it on. An account system means a server. A server means your recordings become something I have to keep up.

The whole design assumes one person on their own machine. Team management is not a feature I have not got to. It is a different product with a different bill.

No meeting bot

yapyap never joins a call as a participant. System audio capture records the call locally, and nothing announces your recording to anybody else's vendor.

A bot is genuinely easier to build. It is also a guest in the room, sitting in a vendor's infrastructure, holding audio from people who never chose that vendor. The custodian this tool is built for — the journalist, the lawyer, the researcher — cannot use it. A bot in a privileged conversation is a third party in a privileged conversation.

So the harder path: capture what your own machine can already hear. Longer to build, works on every platform at once, and adds nobody to the call.

No server of mine holding your data

No sync server, no backup, no archive, no account, no convenience copy.

This is the expensive one, because it is the answer to almost every support request that starts with "could you just". Could you just restore my library. Could you just move it to my new laptop. If your recordings sat on my machine, all of those become one-line answers. They also make me a company holding thousands of hours of other people's privileged conversation, with hosting bills that recur monthly, which is precisely the cost structure that forces a subscription. Say yes here and every other refusal on this page collapses within two years.

There is one bounded exception and it is not live today, so read it as the plan. Some machines cannot run the pipeline at a speed anyone would tolerate. For those I want to rent out hardware that can, charged by the minutes actually spent. Borrowed hardware with a meter on it. Opt in, off by default, per job, processed and deleted. Not a cloud, not a tier, not a plan, not a subscription, and the local path is never gated behind it.

No shared transcript workspace

No real-time collaborative editing. Transcripts are personal documents, not a room other people are typing in.

Multiplayer text means presence, conflict resolution, and a relay to pass the edits through — a relay that would carry transcript content, which is the one thing that never leaves your device without you moving it. You can export a transcript to Markdown and send it wherever you like. The file travels. The archive stays.

The point of a list like this

None of these refusals is permanent by law. I write the software, so I could change my mind about any of them tomorrow.

What makes them worth printing is that they are load-bearing. Each one is holding up the price, and I have written down what would have to break for it to move. If a future version of yapyap ships video, seats, a bot, or a server holding your archive, you will not need me to explain what happened. You will be able to look at this page and see which line went first.