From listening to memory, fully automatic

After listening ends, an automatic pipeline takes over — transcribe, segment, summarize, aggregate, recognize, answer.

Auto-transcribed into golden documents

Every session becomes a transcript — timestamped, speaker-tagged, AI-polished.

100 languages — English, Japanese, Korean, Chinese, German, Italian, French, Spanish, and more
Tap a line to replay — on the device that listened

Search transcript

Mike10:31:12

"The 200ms latency plan — I thought it over, and it still won't fly."

Sarah10:31:48

"Then let's scope the fallback path first."

Possibly Alex · 42%Tag10:32:05

"I can take the client call tomorrow."

Your day, auto-split into activities

Your day, split into six activity types on a timeline.

Meeting · work · meal · social · learning · chit-chat
Activity types recognized from the conversation itself — nothing to pick by hand

2026 / 4

Sat 25 Sun 26 Mon 27 Tue 28 Wed 29 Thu 30

10:30 12:00 MEETING 10:30 – 12:00 · 1h 30m App 2.0 release review — scope, search latency and ship date Scope is locked: search ships in 2.0, offline mode moves to 2.1 — we hold the date. DavidMikeSearch API latencyApp 2.0 review

12:30 13:30 MEAL 12:30 – 13:30 · 1h Lunch with Mike — latency follow-up MikeSearch API latency

14:30 15:00 WORK 14:30 – 15:00 · 30 min Voice note on the walk back App 2.0 review

16:00

17:00 17:30 WORK 17:00 – 17:30 · 30 min Product UI review · Memory palace JennyMikeMemory Palace UI direction

Home Timeline Outputs Agent

The right summary for each activity type

Summaries follow the activity type.

Meetings: structured minutes — agenda, decisions, action items
Meals & chit-chat: a light note — nothing turned into more than it was

MEETINGSarahMikeApp 2.0

App 2.0 release review

10:30 – 12:00 · 1h 30m · Meeting room

Green light: the new search view ships with App 2.0 — beta feedback has been glowing.

✓ Share the beta highlights with Daniel Mike · Fri

Topics aggregated across time

The same thread across days comes together as one topic.

Every related activity attached, most recent first
Add a topic you care about — related history gathers automatically

Q3 TikTok campaign

27 discussions · first 3 weeks ago · latest 2h ago

Luna and Marco keep circling back to creator shortlists and posting cadence.

Creator pitch meeting · Today · 10:30 – 12:00

Client review call · Jul 22 · 15:00 – 15:30

MarketingCampaignMELuna

Voiceprints tell every speaker apart

Voiceprints track speakers across sessions — name once, everything updates.

Confidence hints — "Possibly Alex"
Anonymous until you confirm

About this speaker

Mostly speaks in creator-deal contexts; handles rate negotiations.

"I got the creator down to roughly this rate…"

‹▶›14:32 · Creator call · heard 4×

Yes, it's Marco 78% Someone else Ignore

A built-in Q&A agent

Ask your memory directly — multi-turn, grounded in your transcripts.

A conversation with your memory across time
The same power exposed over MCP

> what did Daniel promise me in last Tuesday's sync?

▸ hemory.search_memory("Daniel", 2026-07-14) · 3 sessions, 14 lines

Daniel committed to the final launch budget by Thursday — and his team owns the App Store screenshots for the Sept 2 launch.

> draft a follow-up checklist

▸ generating checklist.md ✓

Listening on Apple Watch

The watch captures and uploads on its own.

10 hours of listening on Ultra 3 / Series 11 — no more than 30% extra battery usage
On your wrist, closer than any phone — a call pauses listening, and it resumes on its own

Watch listening face: voiceprint dot ring, big clock, listening timer

Connect your agents over MCP

A standard MCP server, with an ever-growing set of tools.

Works with mainstream agents — Claude Code, Codex, Hermes, and more
Unleash your memory's full potential in your agent tools

MCP key

hm_live_••••••••_7f3a

+ New key

Claude CodeCodexCursor+ standard MCP clients

✓ Read-only · bound to this vault

Listening on a schedule, off your mind

Put listening on a schedule — nudged right on time, one tap to start or stop.

Schedule mode: e.g. weekdays 9:00–18:00
Manual mode keeps your privacy boundary

Listening schedule

Weekdays · 09:00 – 18:00ON

WeekendOFF

Listening · 51 min · auto

Platforms

iOS / Apple Watch Get Hemory for iOS

Coming soon macOSWindows / LinuxWeb appAndroid / Android WatchSelf-host

Pricing

Free Trial

$0 one-time

Try it now, no payment.

Listening

10 hrs

one-time · no renewal · VAD*

Everything included

100-language transcription
AI cleaning pass
Auto activity segmentation
Type-aware summaries & minutes
Topics aggregated across time
Voiceprint speaker recognition
Built-in Q&A agent
MCP for your own agents
Apple Watch listening
Scheduled & manual listening
VAD* metering — silence never counts

Starter

$10 /mo

The budget pick — just the conversations that matter.

Listening

20 hrs / mo

VAD*

Everything included

100-language transcription
AI cleaning pass
Auto activity segmentation
Type-aware summaries & minutes
Topics aggregated across time
Voiceprint speaker recognition
Built-in Q&A agent
MCP for your own agents
Apple Watch listening
Scheduled & manual listening
VAD* metering — silence never counts

Pro

$20 /mo

For professional use, enough for everyday work.

Listening

60 hrs / mo

VAD*

Everything included

100-language transcription
AI cleaning pass
Auto activity segmentation
Type-aware summaries & minutes
Topics aggregated across time
Voiceprint speaker recognition
Built-in Q&A agent
MCP for your own agents
Apple Watch listening
Scheduled & manual listening
VAD* metering — silence never counts

Max

Start building your memory palace

Start listening on iPhone and Apple Watch.

Get Hemory for iOS

macOS, Windows, Linux, Web, Android, and Self-host coming soon