Early access — limited spots

AI, already inside
iMessage, your Watch & every tab.

Claude, GPT, Gemini, Grok and Perplexity — right where your conversations already live. One account. No copy-paste. No app switch.

Join the waitlist See how it works
Bring your own keys· End-to-end encrypted keys· Free to start
Apple Watch
3:28
inthread
gpt-5.4
Keyboard
🎤Dictate
Send
11:30 ●●●● ▲ ▮▮▮
SK
Sarah — Client
active now
Need a response to the investor brief by EOD — can you help draft?
+
iMessage
🎤
🔑 Keys
Ask for rewrites, replies, or summaries.
OpenAI
gpt-5.4
Send
Ask AI something…
🎤
Chrome
🔒mail.google.com
OpenAI
gpt-5.4
No API key set. Add one in Setup →
HistoryClear
G OpenAI·gpt-5.4 just now
You
Reply to this Gmail draft — concise + confident
PPerplexity·sonar-pro2h ago
You
What's NVIDIA's latest reported revenue?
$35.1B in Q3 FY25, up 94% YoY [1].
Ask anything… ⌘↩
Send
The hidden tax

Every AI reply costs you
a trip out of the conversation.

Copy the text. Switch apps. Re-explain the context. Wait. Copy the answer back. Switch again. Paste. Edit. Send. You do this dozens of times a day — and it quietly adds up.

0seconds, every time

The round-trip to ChatGPT and back — for a single message you could've answered in three.

Copy Leave app Re-explain Wait Copy back Switch back Paste & fix
0messages a day that need a real reply — a client, a supplier, a recruiter, a DM you can't fumble.
0minlost every single day to copy-paste-switch-paste. Time you'll never see itemized on any bill.
0daysof your year — gone — just shuttling words between two apps.
The detour
~47sec
  1. 1Copy the message you need help with
  2. 2Leave the conversation, open ChatGPT
  3. 3Paste the message, re-explain context
  4. 4Wait, copy the response back
  5. 5Switch back, paste, edit, send
With Inthread
~3sec
  1. 1Tap Inthread, type or dictate the ask
  2. 2Send.

No copy-paste. No app switch. The reply drops straight into the conversation — you never leave it.

The solution

One AI account.
Every surface you talk on.

We brought the models to where conversations already happen — not the other way around. Same five providers. Same eighteen models. Three native apps that look and feel like they belong.

iMessage · iOS

Inside every thread

A native extension that lives in the iMessage keyboard tray. Type, dictate, pick a model, drop the reply into the message field — without leaving the conversation.

Apple Watch · watchOS

On your wrist

The first AI assistant built natively for Apple Watch. Glance, pick a model, dictate, send. Your phone stays in your pocket.

Chrome · macOS / Windows

In every tab

Works on Gmail, LinkedIn, Slack web, Notion, anywhere with text. Page context goes with the prompt, including an optional screenshot for vision models.

Built for both

For you. For your team.

Same engine. Two ways to use it.

Bring your own keys.
Pay zero markup.

If you already pay Anthropic, OpenAI, Google, xAI or Perplexity directly, you can use those same keys here. We're a passthrough — your provider bill is what it always was.

1

Zero-knowledge key sync

On iPhone and Watch your keys stay on the device. Sync to Chrome and they're encrypted with a passphrase only you know (AES-256-GCM) before they ever leave — we store ciphertext we can't decrypt.

2

Web-aware via Perplexity Sonar

Ask for the latest earnings number, news headline, or score. Sonar reads the web and cites sources inline.

3

Voice-first

Dictate prompts on the iPhone, the Watch, or the Chrome popup. Reply hands-free.

9:14●●●● ▲ ▮▮▮
MA
Marco — Supplier
last seen 2m ago
Hola — ¿el envío del lunes está confirmado?
Sí, todo confirmado para el lunes — te aviso cuando salga.
+
iMessage
🎤
🔑 Keys
✦ Anthropic · claude-sonnet-4-6
They're confirming the Monday shipment. Reply: "Sí, todo confirmado para el lunes — te aviso cuando salga."
Anthropic
sonnet-4-6
Send
"Translate and write a polite reply"
🎤

Same voice.
Across every seat.

For agencies, sales teams, and brand-sensitive orgs. Centralized keys, model governance, brand-voice prompts, and per-seat usage analytics.

1

Model governance

Allow or block specific providers and models per team or role. Lock the org to your approved set.

2

Brand-voice prompts

Inject org-wide system prompts that shape every reply. Scope per provider or per model.

3

Centralized keys, AES-256-GCM at rest

One master API key for the team, encrypted in Postgres with a per-row IV — decrypted only in memory to route your team's requests, never logged or exposed.

4

SSO + usage analytics

Clerk-backed SSO (Apple, Google, Microsoft). Per-user, per-model, per-platform reporting.

app.inthread.ai / admin / models
Acme Creative · 24 seats · admin@acme.co
Provider access
Anthropic
opus · sonnet · haiku
OpenAI
gpt-5.4 family
Google
gemini-3 family
xAI
grok-4 family
Perplexity
sonar family
Pricing

Honest pricing.
No surprises.

Free if you bring your own keys. Pro if you'd rather not. Enterprise when you need control.

Free
$0 forever
Bring your own keys

Full Inthread on every surface. Pay your provider direct. We charge nothing.

  • iMessage, Apple Watch, Chrome
  • All 5 providers · BYOK
  • All 18 models
  • SSO via Apple, Google, Microsoft
  • No managed model access
  • No voice dictation included
Join the waitlist
Enterprise
$15/seat/mo
Min. 5 seats · billed annually

Model governance, brand voice, analytics. For teams that need to ship the same voice.

  • Everything in Pro
  • Model governance per role
  • Brand-voice system prompts
  • Centralized keys (AES-256-GCM)
  • Usage analytics & cost reporting
  • SSO + SCIM provisioning
Contact sales

Be first to use
AI in your thread.

Opening access in waves. Drop your email — iMessage, Watch and Chrome ship together.

No spam. Three platforms, one signup.
You're on the list. We'll be in touch soon.