Short answers.

What is Lorah?

Lorah is a secure, local-first AI workspace. You run a team of persistent AI specialists (“teammates”) that share project memory — you approve what becomes shared truth. Your workspace, your keys, your machine. Multi-model: OpenAI, Anthropic, Gemini, Perplexity, xAI (Grok). macOS and Windows.

Learn more →

Who is Lorah for?

Founders, consultants, operators, researchers, and AI power users who already use multiple AI tools and lose time re-explaining project context. If you flip between ChatGPT, Claude, Gemini, and Perplexity for the same project, Lorah is the workspace around them that keeps the context.

How is Lorah different from ChatGPT Projects or Claude Projects?

Four things:

Local-first. Your workspace lives on your machine, not in their cloud.

Multi-provider in one workspace. Lorah spans OpenAI, Anthropic, Gemini, Perplexity, and xAI (Grok) — you don’t pick a platform-lock.

BYOK with zero markup. You bring your own API keys and pay the provider directly.

Reviewed memory with provenance. You decide what becomes durable memory. Every memory entry traces back to the conversation that produced it. ChatGPT and Claude Projects capture memory automatically and silently.

Can I use Lorah with Claude, ChatGPT, Gemini, and Perplexity at the same time?

Yes. One workspace, five providers. Each teammate is bound to one provider at a time, but a single workspace can have teammates on different providers running side by side. Rooms let them work on the same question from different models.

What does “local-first” mean?

Your workspace data — teammates, memory, conversations, library — lives on your machine. Lorah’s servers do not host any of it. There is no “Lorah cloud” mirroring your data across devices, because we don’t have one. Network is only used for model calls (direct to providers) and license / update checks.

Full trust story →

Does Lorah store my workspace data?

Lorah’s servers don’t store it. Your workspace lives on disk where you installed Lorah. Our servers see two things: licensing (email, Stripe customer ID, plan type, status, date) and a daily license check-in that carries a random install id, the app version, your operating system, your license type, and an approximate country — never an IP address. No logging of prompts or completions. Crash reporting is off by default. If you turn it on, at first run or in Settings → Data & Privacy, Lorah sends short content-free reports: the error type, your app version and platform, and whether a sync succeeded or failed. Never chat content, memory, file names, prompts or API keys. You can read every queued report in Settings before it’s sent. Two further switches, also off by default: a redacted diagnostic snapshot, and linking reports to your install’s license.

Full trust story →

How does Lorah use my API keys?

You bring your own keys (BYOK). They’re stored locally, encrypted via your operating system’s keychain (AES-256-GCM). When a teammate makes a model call, the request goes from your machine directly to the provider. Lorah’s servers are not involved, and Lorah adds zero markup on inference — you pay the provider at their published rate.

What’s a “teammate”?

A persistent AI specialist with a role, prompt context, and accumulated memory. Long-lived, not session-based. You give it a name, a personality, a mandate. It builds memory across every conversation. Common starter teammates: Strategist, Builder, Researcher, Critic, Writer.

What are Rooms?

A room is where you ask the whole team one question at once. Pick the people, ask, and each of them answers on their own, side by side, so a Strategist, a Builder, and a Critic can weigh in without four chat tabs.

When you want a verdict, press Reach a decision and Lorah writes where the room landed, disagreements first. If you would rather have them work turn by turn, Work together opens a working room instead.

Rooms walkthrough →

How does Lorah memory work?

Memory in Lorah is reviewed, not automatic. Teammates propose memory entries during conversation. Proposed entries land in the Review Queue. You approve, edit, or reject each one. Approved entries become durable memory with provenance (a link back to the conversation that produced it). You can supersede entries when facts change, or reject them entirely.

This is different from automatic chat-history memory in other tools, where everything is remembered silently and nothing is traceable.

Is Lorah an autonomous agent platform?

No. Lorah is workdesk-style, not agent-style. Important outputs — memory, sync briefs, MCP write-backs, task results — pass through human approval in the Review Queue. Lorah is designed for serious work that benefits from oversight, not for unattended long-running action.

Can Lorah connect to Claude Desktop, Cursor, or Claude Code?

Yes. Lorah supports MCP. Lorah can act as an MCP server so compatible tools can read approved workspace context, and Lorah can connect to external MCP tools through teammates.

Write-back goes through the Review Queue before it becomes durable memory.

MCP setup guide →

Why use Lorah next to an assistant I already have?

Use Lorah when a project outlasts one chat. Your teammates’ roles, your materials and the memory you approved live with the project, so you can switch models without setting everything up again.

For a quick question, one assistant is enough. Lorah is for work you will still be doing next week.

Does review make an answer correct?

No. Review means you accepted it into project memory. It records your decision, not a fact check.

You can always see where an item came from, and replace it later when the project changes.

Can I start without the Compiler?

Yes. Pick a starter workspace, or build a team by hand.

The Compiler needs an Anthropic or OpenAI key, or a local model that qualifies. Gemini, Perplexity and xAI keys work for teammates, not for the Compiler.

Can I use my ChatGPT or Claude subscription?

No. Lorah works with API keys. A key is separate from a chat subscription and is billed by usage, by the provider.

Creating a key takes a minute on the provider’s site. Lorah adds no markup.

Does a local model make Lorah offline?

Your chats with that teammate stay on your computer. Sync, decisions and Ask Lorah run on the Compiler’s runtime, which can also be a local model.

License checks, updates and web search still use the internet.

What happens when Pro expires?

Nothing is deleted. Chat and Ask Lorah keep working with 3 active teammates.

Sync, Remember this, new Rooms, background tasks and MCP pause until you upgrade. Memory, materials and history stay readable.