LookinGlass docs

Three minds, one answer. The shortest path from "I have a question" to "I have an answer I trust."

Overview

LookinGlass is multi-model AI chat. Instead of being locked to a single AI, you can ask the same question of multiple frontier models, compare their reasoning side-by-side, and let a synthesis layer fold the best of each into a single answer. Same model ecosystem you already know, framed so disagreement becomes obvious rather than hidden.

Available on:

• iOS — universal app on the App Store.
• Web — lookinglass.ch, same account, same history.

Sign in

Two ways:

• Email & password. Use the credentials your admin set for you.
• Fresnaye SSO. If your organization uses Fresnaye SSO, tap "Continue with Fresnaye SSO" on the login screen.

Don't have an account? Request access — we'll get back to you.

Your first chat

Tap + new chat. Type a question. Hit send. By default you'll be in Standard mode talking to Claude. To compare two models side-by-side, swipe to Parallel before sending. To get a single synthesized answer from all three, use Bash. To see where the models agree and where they disagree, use Triangulate.

The models

LookinGlass runs three frontier models by default, plus Fable 5 in private beta. Each is color-coded so Parallel mode stays readable at a glance.

Standard, Parallel, Bash, Triangulate

Picking a model

There are three ways to control which model you talk to:

1. Default preference. Set your preferred default in your profile. Every new chat starts there unless you override.
2. Per-conversation lock. When you start a new chat, lock it to a single model. The whole conversation stays on that model, even if your default changes.
3. Workspace default. Your workspace admin can set a default model for everyone in your organization.

Precedence (most specific wins): conversation lock → user default → workspace default → system default.

Projects

Projects let you group related chats and attach a custom system prompt. Every chat inside a project inherits the project's rules — handy for "always answer in formal tone," "default to USD currency," or domain-specific style guides.

Create a project from the drawer. Move existing chats in by long-pressing them.

Document chat

Drop in a PDF, Word doc, Excel spreadsheet, or PowerPoint deck and ask about it. We extract the text for search and feed the pages (rasterized for PDFs and Office files) directly to the model so it can read tables, layouts, charts, and embedded images.

HEIC photos from iOS are auto-converted to JPEG so the models read them cleanly.

Memory

LookinGlass auto-remembers facts about you across chats — your name, role, preferences, recurring projects. You can also add memories manually from the profile screen. Every new turn includes your pinned memories in the system prompt so the model has context without you having to repeat yourself.

Search

Use the search bar in the drawer to find a phrase across every chat you've ever had. We index conversation content plus your memories. Matches show snippets with the keyword highlighted; tap one to jump straight to that turn.

Privacy & data

Your conversations are stored on our servers and routed to the underlying provider for each turn. We use the API tiers that contractually exclude your data from training. We don't share your data with advertisers, and we don't browse user content casually.

For the full policy see our privacy policy. For deletion, export, or specific questions, email support@fresnaye.com.

Releases

See the changelog for what's new.