rAlph – Slash commands (/help, /report, /drafts)

rAlph is only available when AI is enabled for the tenant. Without AI enabled, neither the Assistant entry in the sidebar nor the floating chat button appears. Administrators enable AI under Settings → System & Automation → AI Settings.

You consciously release the data rAlph may see. Under Settings → System & Automation → AI Settings → Allowed services you decide which areas (e.g. People, Roles, Value streams) and services (Aggregator, Validator, …) the AI may process. If an area is not released, the AI does not process its data — neither in the chat nor for semantic search (embeddings). By default nothing is released, so you actively grant access. This keeps you in control of which organizational data reaches the AI.

Does rAlph “not find” something that clearly exists? Then the relevant area is probably not released. During search rAlph actively points this out when an area (e.g. Value streams) is blocked by the AI releases — instead of simply reporting “nothing found”. An administrator releases the area under Settings → System & Automation → AI Settings → Allowed services, after which rAlph finds the data.

Type a ”/” in the chat input to open a command menu. It lets you start focused flows without long phrasing. Use the arrow keys to choose, Enter or a click selects the command. A chip then appears above the input showing the active command, which you can remove anytime via the ×.

/help – questions about using the app

/help <your question> answers questions about how to use and operate roleALPHA. rAlph searches the help pages and bases its answer on the sections it finds, including links to the relevant pages on doc.rolealpha.app. Example: “/help How do I create a relationship?”

/help is read-only – it never changes your data.

/report – create an analysis

/report <description> (or /aggregator) creates a new, saved report. Just describe what you want to analyze, e.g. “/report number of roles per cost center as a bar chart”. rAlph creates the report (owned by you, visible only to you), runs it and shows the chart right in the chat. You’ll then find the saved report on the Aggregator page.

You don’t strictly need to type the command: if you ask rAlph for a chart in a normal chat (“Create a bar chart of all my roles”, “How many people per cost center?”), it also creates a report automatically and shows the chart inline.

Add as a dashboard tile: below the chart you’ll find “Add to dashboard”. One click places the report as a selection tile on your personal dashboard, where it will be shown to you directly from then on.

This command is only available to users with the required permissions (tenant or platform admin) – only then does it appear in the menu. In addition, the Aggregator service must be enabled for the tenant under Settings → System & Automation → AI Settings so that rAlph is allowed to create reports.

/drafts – approvals right in the chat

/drafts shows the open drafts you need to act on – as a list right in the chat, so you don’t have to look up each draft individually. Depending on the approval mode you can, per entry:

Approve or Reject (when you are the approver), or
submit an agree/object decision (consensus/consent).

Use “Approve all I’m allowed to” to handle several approvals at once.

When rAlph creates drafts itself (e.g. relationships in Plan mode), it always shows them as a draft card right in the chat — with a link to the drafts area. The cards appear directly under the answer that proposed them: as you continue the conversation, each suggestion group stays with its message (instead of piling up at the bottom of the chat), and “Create all” applies to exactly that group. So you see immediately what was proposed in which step. rAlph cannot delete entities: start a deletion via the entity’s own delete action (which creates an approval-gated deletion draft). Details: Drafts.

/automator & /validator – create rules (admins only)

Use /automator to describe an automation in words, e.g. “Whenever a circle is created, create the LeadLink role and link it to the circle.” Use /validator for a validation rule, e.g. “A person must not be attached to more than one cost center.” rAlph shows a rule proposal as a card; only your click on “Create rule” creates it. Both commands appear only for users with admin rights. Details: Automator, Validator.

/insight – cross-domain questions

With /insight rAlph answers questions that span several areas and their relationships — e.g. “Which roles deal with projects?”, “Which roles and value streams work competitively?” or “Where do content and relationships not match?”. Details: Cross-domain questions (graph + semantics).

/analytics – usage numbers for your tenant

With /analytics (also /usage) rAlph answers questions about your tenant’s product usage — e.g. “How is our team using the app this week?”, “Which modules are edited most?” or “How many entities were created?”. The numbers are strictly limited to your tenant (other tenants are out of reach for rAlph) and aggregate/pseudonymous (no personal data). Available to all users when AI is enabled. Backed by Usage Analytics.

/context – view & fill the AI context from a website/document (admins only)

The AI company context (business model, industry, company size, culture, strategic goals) feeds into every answer rAlph gives. So far you maintain it manually under Settings → System & Automation → AI Settings. With /context you can do it right in the chat:

View: /context shows you the currently stored context in readable form.
Fill from a website: Provide an address, e.g. “/context Fill the context from https://www.my-company.com”. rAlph reads the public page, derives the matching fields and fills them in.
Fill from a document: Attach a document (e.g. a PDF with your company profile) and write “/context Fill the context from this document”.

rAlph writes only fields that actually change and summarizes the changes at the end. The command is available to admins only – only they may change the shared company context.

/agent-context – validate the external agent view (admins only, module required)

With /agent-context <agent> rAlph shows an AI agent’s context bundle (roles, value streams, competences, plus the binding policy guardrails) exactly as an external agent would receive it over its own API key — unfiltered, not your own view. This lets you check right in the chat what an externally connected agent actually gets as context, without switching to the agent’s detail page. It’s the same view as “View as agent” in the debug area of the agent detail page (see AI Agents) — just directly in the conversation, where you can then ask follow-up questions about this exact bundle (e.g. “does this guardrail also apply to action X?”) without rAlph having to pull it again.

This command appears only when the AI Agents module is enabled for the tenant, and only for users with admin rights (tenant or platform admin) — the unfiltered view is deliberately not available to everyone. In addition, the AI Agents service must be enabled for the tenant under Settings → System & Automation → AI Settings.

Reading web pages – anytime

Independently of /context, you can put a URL into your message anytime or ask rAlph to read a page – e.g. “Summarize https://…”. rAlph then automatically reads the web page and bases its answer on the page content. Only public pages are read (no JavaScript rendering, no internal/private addresses).

Important: rAlph never decides on its own

rAlph never approves anything itself. With /drafts it only presents the drafts – the actual approval is always your click and goes through the normal approval path (see Approval Modes). And when you ask rAlph to change entities or relationships, the result is always a draft – never a direct live change. Nothing takes effect without your confirmation.

Stopping a response

While rAlph is answering, the send button turns into a stop button (red square). Clicking it interrupts the running response immediately — the partial answer written so far stays, and generation is stopped on the server too (no further tokens are spent). Unlike “New chat”, your existing history is preserved; you can send a new message right away.

Switching tenants: a fresh session

When you switch the tenant via the tenant switcher at the top, rAlph starts a fresh session: the previous chat history, pending suggestions and scenario selection are cleared. This keeps another tenant’s context from carrying over into the new one. At the same time, your permissions and the navigation follow the selected tenant from that point on – if you are, say, an administrator in one tenant but only a viewer in another, exactly those rights apply. You can only change a tenant’s data from its active session.