> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.rydhm.ai/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.rydhm.ai/_mcp/server.

## Rydhm

The product is now Rydhm, with a new look across the app, the sign-in pages and account emails. Experts show their photo or a colour orb.

## Developers section

A new **Developers** section for workspace admins:

* **API keys** with scopes, optional limits to some experts and to IP addresses, expiry dates, and rotation with a grace period. See [Authentication](/developers/authentication).
* **Service accounts**, so keys don't depend on any one person.
* **Request log** of every API request made with your keys.
* **Webhooks**: signed events for conversations, calls, meetings, summaries, knowledge, tools and experts, with retries, a delivery log and test events. See [Webhooks](/developers/webhooks).
* Calls your experts make to your custom tools are now signed the same way, so you can check they came from Rydhm.

## Workspace

* **Settings** now covers members and invitations, workspace policies, API key rules and data retention.
* An **audit log** of who changed what, with filters and CSV export.
* **Alerts** and a notifications bell, with settings for what reaches you.
* Sign out automatically after a stretch without activity.
* **Allowed websites:** choose which sites can embed an expert, per expert or for the whole workspace. See [Embed](/developers/embed).

## Building experts

* **Describe it:** describe the expert you want, and Rydhm researches and writes it for you. Or start from a template.
* Attach documents while you create an expert.
* Give an expert a colour or gradient background instead of an image.
* **Web search:** experts can search the web during voice calls and chats. It is off by default; turn it on per expert, and optionally limit it to some sites.
* **Coach:** an AI coach on the expert's page. It looks at the expert's setup and conversations, suggests changes, and rehearses them in simulated conversations before you apply them.

## Tools

Tools are easier to understand and to trust: each tool shows its health and recent calls, you can **Try it** before saving, and a meeting's transcript shows the tool calls made during it. See [Tools](/experts/tools).

## Voice

* **GPT-Live** is a third voice engine for 1:1 calls, with its voices in the voice library. See [Voice engines](/experts/voice-engines).
* On 1:1 calls, the expert stops the moment you start talking, and carries on after a short "mm-hmm".
* **Voice controls** per expert: speed, tone, pronunciations and a backup voice.
* **Deep thinker:** a stronger model the expert can consult for hard questions, such as comparisons and calculations, during a call. Off by default.
* A **thinking sound** while the expert works on an answer.
* In meetings, experts no longer stop at a "mm-hmm", handle hesitant speakers better, and read numbers back carefully.
* The expert's tile in a meeting is now a live orb that moves with its voice.