Why We're Building Sayforth

Sayforth is a 30-day speaking program for iPhone. You answer one question out loud each day, for about a minute. The app draws your answer as a single line, tells you one thing that happened, and gives you one change to carry into tomorrow. That's the whole loop. No dashboard, no library to browse, no decision to make about what to do today beyond starting.

Most people find it because an interview is coming up. That's the honest entry point: a date on the calendar makes practice feel urgent in a way that "get better at meetings" never quite does on its own. But the skill underneath is the same one you need in a meeting where you go quiet, hedge everything, or explain for two minutes before you get to the point. So the program works for both, and you choose which one you're aiming at when you start.

Why there's no score

Every speaking app we looked at before building this one gives you a number. A confidence score, an energy score, a filler-word count, something out of 100. We didn't build that, and it wasn't an oversight.

A score answers a question nobody actually asked. You don't want to know that you're a 74. You want to know what happened in this specific answer, and what to do differently in the next one. So instead of a score, Sayforth shows you a line: a stroke for the part where you were talking, a gap for a pause, sized to how long the pause actually was, and a heavier stretch where your main point landed. Underneath it, one sentence about what happened, one change for tomorrow, and a short drill if you want to try it right now. That's the whole feedback screen. It takes about five seconds to read because there's nothing to compute: you can just look at it.

The comparison is also different. Sayforth never measures you against other people. There's no leaderboard, no percentile, no "better than 80% of users." Day 1 is a baseline you set for yourself, and everything after that is measured against it. On Day 30 you answer the same question you answered on Day 1, and you hear both, side by side. That's the only comparison that means anything: you, thirty days apart.

Why your recordings stay on your phone

Your voice is a strange thing to hand over to a stranger's server by default, so we didn't build it that way. Timing and speech-to-text happen on your iPhone. What leaves the phone, when you're on a coached session, is the text of what you said and some numbers about it (pace, pause length, where things landed), not the audio itself. That's the whole privacy story for the current build: no audio upload, transcript and metrics only, and only when a coaching read is actually happening.

This wasn't a hard call to make. It was harder to build than the alternative, and it means we can't do some things a cloud-audio approach would let us do more cheaply. We think that trade is worth it for something this personal.

Where the project actually stands

As of today, the app is built and in testing on real devices. We're preparing the App Store submission: the listing copy, the privacy disclosures, the screenshots, the review notes, all of it. The app is not live yet, and I'm not going to put a date on this post that I might have to walk back later. This website is live at sayforth.app, ahead of the app itself, partly so there's somewhere honest to point people who hear about this before it's downloadable, and partly because writing the pricing page and the privacy page before the app ships is a good forcing function for making sure both are actually true.

The 30-day structure is built: four weekly phases, a Day 1 baseline, daily coached reps, evidence check-ins along the way, and a Day 30 retest you can hear. Coaching feedback is generated from your transcript and your timing numbers, never your audio. Pricing is settled too: free to start, with the full 30 days and its own coaching on your first few sessions and then periodically after. Daily coaching, which keeps every session coached, is $9.99 a month or $59.99 a year with a 7-day trial on the annual plan. No weekly plan, no forced trial, no paywall before you've even tried it once.

What comes next

Submission first, then whatever Apple's review turns up. We'd rather find problems there than after people are using it. After that, the plan is to keep this build log honest: what we changed, what testing told us, what we cut. If you want the short version of what the app actually looks and feels like before it's out, a full day, step by step, is on the site.

We'll write here as the real milestones happen, not on a schedule. The next honest thing to report is submission.