Saysort

Saysort · Voice inbox for Hermes Agent

Say it once.
It comes back sorted.

Saysort is an iPhone and iPad app that puts a microphone on your own Hermes Agent. Talk the way you think. It hears your names and jargon spelled your way, and files what you said as tasks, notes and messages in the right project.

Email me at launch How it works

Coming soon to iPhone and iPad · in private TestFlight today

TUE 08:14 · WALKING · 0:19 “Send Sam the Q3 numbers before Friday. Oh, and for the pricing page: one plan, no tiers. Tell Lina the site review moves to Thursday.”

Task

Send Q3 numbers to Samdue Fri · → Finance

Note

Pricing page: one plan, no tiers→ Website

Message

To Lina: site review moves to Thursdaysent by your agent · → Website

Illustration. Your agent picks the project. Wrong? Say “no, that’s Marketing.”

Why

Your best thinking happens away from the keyboard.

Without it

  • Voice memos you never play back.
  • One long chat with your agent that mixes ten topics.
  • Dictation that turns the names you use every day into nonsense.

With Saysort

  • Each thing you say lands in its project, tagged so you can see where it went.
  • Your vocabulary: people, products and code words come out spelled right, from the same dictionary your agent uses.
  • One glance shows what’s moving, what’s stuck and what’s waiting on you.

How it works

Pair once. Then just talk.

Pair with your agent

Open one pairing link from your Hermes Agent. The server address and token go into the iOS Keychain. You don’t make an account with us.

Talk

Tap the mic, or turn on hands-free and leave the phone in your pocket. Cut in at any time: it stops speaking and listens.

Get it back sorted

Every message gets a project chip, like “→ Website”. Your agent answers out loud. Questions it needs you to decide arrive as one-tap buttons.

In the app today

Built for talking, not typing.

Talk
A big mic and a live transcript. Hands-free mode sends when you stop talking and listens again after the reply. Works with AirPods and with the screen locked.
Two voices
Standard uses speech recognition on your device and spoken replies from your agent. Real-time voice is an optional back-and-forth conversation mode, if your agent is set up for it.
Your words
Transcripts use your agent’s dictionary. Add a name or a correction once and it is spelled right from then on, without an app update.
Pulse
Your projects in four groups: Moving, Stuck, Waiting on you, Done. Each has its own timeline, and Saysort can read it aloud.
Decide
The questions your agent is waiting on, each with one button per answer.
iPad
A real iPad layout, with the transcript and cards side by side in both orientations.
Demo mode
Try every screen before you connect anything. Demo mode runs on sample data, on the device.

In development

  • Briefs inbox
  • Full to-dos
  • Past conversations
  • Scheduled jobs
  • Files
  • Skills
  • Memory
  • Model switching
  • Notifications

Listed so you know where it’s heading. These are not in the TestFlight build yet.

The honest part

It’s the voice front end for an agent you already run.

What you need

An iPhone or iPad on iOS 18 or later, and your own Hermes Agent that your phone can reach. Saysort doesn’t include an AI model. Your agent does the thinking, with the models and keys you chose.

What we don’t do

No account with us, no ads, no analytics, no third-party SDKs. We don’t run a server between you and your agent. Details are in the privacy policy.

Not on the App Store yet.

Email me at launchQuestions