Skip to content
The world does revolve around you.

What to do next#

Your Hub is installed, the AI model is loaded, and at v1.0 the instant-onboarding step has done the heavy lifting up front: iMessage, WhatsApp (if you opted in), Safari and Chrome history, Apple Mail, Notes, Calendar, and the other sources you opted into are already in your knowledge graph at their full historical depth. The iOS app is paired, or will be soon.

This page is the short list of high-value things to try in your first half hour. None of them take more than a few minutes.

1. Open the Doctor dashboard#

Open your browser to:

http://localhost:8089/doctor

Doctor is your local diagnostics panel. It shows:

  • Service health – database, AI engine, embedding service, import pipeline.
  • Import status – how many people, conversations, events, places are in your graph.
  • Pairing – any iOS apps paired to this Hub.
  • Recent activity – what the Hub has been doing.

If anything is red or amber, Doctor usually has a copy-paste fix command next to the warning. Bookmark this URL.

2. Ask the assistant something you can verify#

The fastest way to feel what Ostler is for is to ask it something you already know the answer to. Open whatever channel you have wired up – iMessage, WhatsApp, or the iOS app – and ask your assistant a small, specific, personal question. For example:

  • "When did I last see [a friend's first name]?"
  • "What did I and [a colleague] talk about last month?"
  • "Who do I have a meeting with tomorrow?"
  • "Find [a person] – what's their phone number?"
  • "Remind me what I owe [someone] a reply on."

Verifiable questions teach you what your assistant is good at on day one. The deeper questions ("what patterns do you see in my work calendar over the last six months?") get more useful as more data flows in.

Calling your assistant by its name

Your assistant only responds when it is addressed directly. In a group chat, that means starting your message with the name you gave it: "Samantha, when did I last see James?". One-to-one, every message is for it. See reference > configuration for how to change the wake-name later.

3. Request your data exports#

The instant-onboarding data already in your graph is just the start. The deeper picture – everyone you have ever connected with on LinkedIn, every Facebook friend, your Twitter / X archive, your full Gmail content – comes from GDPR exports the platforms send by email.

Request them now while you do other things; they take 1 to 3 days to arrive. The full step-by-step is on the marketing site at ostler.ai/getting-started; the short list is:

Platform Where to request
LinkedIn linkedin.com/mypreferences/d/download-my-data – pick the larger archive
Facebook accountscenter.facebook.com/info_and_permissions/dyi – select JSON, low media quality, all time
Instagram Same Accounts Center URL, pick Instagram
Twitter / X x.com/settings/download_your_data
Google (Gmail) takeout.google.com – Deselect all first, then tick Mail only, ZIP, 50 GB chunks
WhatsApp In the WhatsApp app: Settings > Account > Request account info > Account report

Save the ZIPs to ~/Downloads. The Hub has a watcher that scans ~/Downloads every four hours and sends you a notification when new exports show up. When you are ready to import, run:

ostler-import ~/Downloads/ --verbose

The importer is happy with whatever has arrived; it skips what is missing and you can re-run it whenever a new export lands.

4. Plug in Apple Mail#

If you did not already, add your email accounts to Apple Mail:

System Settings > Internet Accounts > add your account > tick Mail.

Once Mail has synced your messages locally (this can take an hour or two for a busy mailbox), re-run the FDA extraction so Ostler picks up the new content:

ostler-fda

Apple handles authentication, message storage, and OAuth refresh. Ostler reads from the local Mail store via Full Disk Access. Google, Microsoft, and Apple never see Ostler exists. See Apple Mail FDA vs Google OAuth for why we built it this way.

The same trick works for Calendar: System Settings > Internet Accounts > add account > tick Calendars. Ostler reads them all together via the local CalDAV store.

5. Plug in WhatsApp (optional)#

If you want WhatsApp messaging through your assistant, the Doctor dashboard offers a WhatsApp Web pairing flow. The Hub uses your existing WhatsApp account – there is no separate Ostler number needed. Once paired:

  • Your assistant can read recent WhatsApp messages.
  • You can talk to your assistant directly via WhatsApp (handy when you are out of the house).
  • Group chats: your assistant only responds when its name is mentioned.

WhatsApp Web is unofficial – read this before you turn it on

WhatsApp does not publish an official API for personal accounts. The Hub talks to WhatsApp through the same WhatsApp Web protocol your browser uses, which is reverse-engineered. That has consequences: Meta can ban the number you use without notice, the protocol can change in ways that break the Hub's connection, and using a personal number with an unofficial client is technically against WhatsApp's terms of service.

Setup in v1.0 surfaces this with an inline tickbox you have to acknowledge before the connector starts:

I understand WhatsApp Web is unofficial; my number may be banned by Meta; I accept this risk.

Your acknowledgement is recorded in ~/.ostler/posture/consent.json so it is auditable later. You can rescind it any time by toggling the connector off in Doctor – the Hub stops talking to WhatsApp immediately.

Like Apple Mail, this is opt-in and can be turned off any time from Doctor.

6. Set up the morning brief and evening wrap (optional)#

Ostler can send you two short summaries each day:

  • A morning brief at 9 am with what is on your calendar today, people you have not heard from in a while, threads you owe a reply on, and birthdays or anniversaries this week.
  • An evening wrap at 6 pm with what got captured during the day and a heads-up on tomorrow.

To turn either on, open Doctor and tick Morning brief or Evening wrap in the Notifications panel. Pick a delivery channel (WhatsApp, iMessage, or email) and a time – 9 am and 6 pm are the defaults.

Both briefs are generated locally on your Hub. They do not leave your Hub except via the channel you choose.

Fail-loud, not fail-silent

These cron jobs are designed to default to best_effort=false so that delivery failures – the channel is offline, the model is busy, the WhatsApp connector is unauthorised – surface in the Doctor dashboard's Notifications panel rather than being swallowed as quiet warnings. The v1.0 implementation is currently best-effort while the fail-loud surface lands in a near-term update; you may need to spot-check the Notifications panel on the first few days of use until that ships.

7. Pair the iOS app#

If you have not already, this is a good moment. The iOS app is your phone-shaped window into your knowledge graph and the place quick captures (snap a photo, paste a URL, jot a note) flow back into the Hub.

Walkthrough: pairing the iOS app.

8. Have a poke around#

Some less obvious things to try:

  • The Doctor's What Ostler knows view – a structured summary of every category of data the Hub holds about you. A useful sanity check.
  • The personal wiki – a generated wiki of every person, place, organisation, and topic in your graph. Available from the Doctor home screen.
  • The timeline – every meeting, message, and event chronologically. Available in the iOS app and from Doctor.

What you do not need to do#

A few things people coming from cloud-AI products expect to do, and don't have to:

  • Sign into anything. No Ostler account, no API key, no telemetry consent flow.
  • Pick a model. The installer chose one for your hardware. You can change it later if you want.
  • Configure prompts. Your assistant has sensible defaults baked in. Most users never touch them.
  • Worry about pricing tiers. The Hub is a one-off purchase; the iOS app is a separate App Store subscription. See pricing on the marketing site.

When something looks wrong#

Open Doctor first. It will usually have already detected the problem and have a copy-paste command waiting.

If Doctor does not help, check troubleshooting. The common ones – Docker not running, Ollama not responding, FDA not granted, ostler-import not on PATH – are all covered there.

If you still cannot work out what is wrong, email [email protected]. Open the Doctor dashboard at http://localhost:8089/doctor, screenshot the Health panel, and attach a copy of ~/.ostler/logs/ostler.log (the recent tail is usually enough, and contains no personal data – you can preview it before sending).

A reasonable expectation for week one#

  • Day one: instant-onboarding data is searchable at full historical depth. iMessage and WhatsApp threads (subject to which you opted into), years of browser history, Apple Mail, Calendar, Contacts, Notes. The assistant has real context to work with, not just the last week of activity.
  • Day two or three: GDPR exports start arriving and extend the picture into the platforms that live outside Apple's walled garden (LinkedIn, Facebook, Instagram, Twitter / X, Spotify, others). LinkedIn, Twitter / X and Google Mail take an hour or two to import; Facebook and Instagram can take a day.
  • End of week one: patterns start showing up. "You haven't replied to this person." "You've talked over them in meetings." The bits that nobody else has.
  • End of month one: the morning brief stops being a curiosity and starts being part of your routine. The iOS app is the first thing you reach for when a name is on the tip of your tongue.

If something on that timeline is not happening, ask in the support email and we will dig in.

Where next#