User Guide

Writing Effective Queries

The quality of Sevrel's answers depends heavily on how you phrase your questions. Here are proven patterns that consistently produce accurate, useful results.

Be Specific About What You Need

Include Context

The more context you provide, the better Sevrel can narrow down the right documents:

• Property name: “at Cedar Point Center”
• Tenant name: “for Acme Coffee Co.” or “for tenant ABC Corp”
• Time period: “for Q4 2025” or “in the 2025 budget”
• Document type: “from the rent roll” or “in the lease amendment”

Query Examples for CRE

Understanding AI Responses

Sevrel responses can include several components designed to help you investigate the result and verify material claims.

Source Citations

Numbered references (like 1) link claims to specific documents. Click any citation to see:

• The source file name, plus its section and page where available
• A relevant excerpt from the document
• Buttons to preview the file in Sevrel or open it in the Documents browser, for sources in your connected storage provider

Thinking Indicator

While Sevrel processes your query, you'll see a real-time indicator showing what it's doing: searching for documents, reading specific files, and synthesizing an answer. This transparency helps you understand how the response was constructed.

Follow-Up Questions

Conversations maintain context. After receiving an answer, you can ask follow-up questions without repeating property names or tenant details. For example:

Sevrel remembers the property and tenant context from your previous question.

How Sevrel Picks an AI Model

Sevrel uses routing heuristics to select a configured tier from our enterprise AI provider. The selection is automatic, but it is not a guarantee of answer quality:

Every interactive chat turn is answered by the Standard tier or deeper. The Fast tier is reserved for the router itself and for supporting work — composer hints, suggestion chips, notification previews, and background jobs — so it never writes the answer to your question.

Routing happens transparently based on inferred query complexity. Framing a question as an analytical request (“analyze”, “compare”, “interpret”) tends to invoke a deeper-reasoning tier, but the classifier or selected model can be wrong. Verify material output. Learn more about the AI models →

Managing Conversations

Sevrel saves your conversation history so you can return to previous research:

• View past conversations — from the sidebar, browse your chat history and filter it by conversation title. The row menu offers Rename and Delete
• Start a new conversation — clear the context for a fresh topic
• Share conversations — from the composer's “+” menu, generate a read-only snapshot link (labelled Live link in the app). The link opens without sign-in, so treat it as public; it expires after 30 days and can be revoked at any time. Incognito conversations cannot be shared

Tip: Start a new conversation when switching to a completely different property or topic. This keeps context clean and improves result quality.

Prompt Shortcuts

The chat catalog contains editable one-turn prompt shortcuts for common commercial real estate questions. Choosing one inserts ordinary prompt guidance; it does not invoke a separate, deterministic pipeline, guarantee an artifact, or start continuous monitoring.

Open the composer's “+” menu and choose Prompt shortcuts to browse them, or type a shortcut's slash command directly (for example /ask-portfolio which leases expire in 2027?). Selecting from the menu inserts just the /slug token — type your own request after it before sending. Typing / on its own does not open a picker. The historical dedicated template panel, 19-template inventory, and custom-template save API are not currently mounted.

Proactive Agents

Sevrel's proactive-agent catalog contains three configurable workflows:

Structured findings appear in the in-app Notifications feed and later briefings when proactive processing is enabled.

Portfolio & Work Orders

The legacy Dashboard route has been retired. Current operational work is split between the Portfolio and Work Orders workspaces:

• Portfolio — confirmed properties, tenancies, lease dates, occupancy, and related source-document context
• Work Orders — maintenance intake, assignment, status, and operational reporting
• Assistant — conversational analysis; proactive findings are reviewed from Notifications

Work Orders

Sevrel includes a work-order module for property operations. Create and track maintenance and service requests directly from chat (“create a work order for the HVAC issue at Building B”) or from the Work Orders view.

• Permission-gated — creating or modifying a work order requires the workorders:write permission on the user's role; viewers can read but never create. The gate is enforced both when the request is staged and again at confirmation time.
• Action policy — supported external actions are evaluated under the organization's server-enforced allow, require-confirmation, or block policy. Chat-proposed work orders are staged for explicit confirmation before creation.
• Operational history — status changes, assignments, and source references are recorded. Selected security and product events are also written to an append-only audit log.
• Tenant intake (early access) — an organization-gated tenant portal lets tenants submit photo-attached requests through expiring magic links, with per-organization isolation enforced at the database layer. See the roadmap for general-availability status.

Exporting & Sharing

You can download the current conversation as a formatted PDF from the composer's “+” menu under Share, or copy it as plain text. PDF is the only conversation export format wired to the UI today.

• PDF — formatted conversation export with a source-citation table, generated in your browser
• Generated files (.pdf, .xlsx) — produced by the in-chat file-generation tools and delivered as a download card in the conversation. New documents are generated as PDF; a .docx comes back only when you ask Sevrel to edit a Word file you already uploaded. These tools are off by default — creating new files needs the admin setting labelled “PDF Generation”, and editing an existing spreadsheet needs either that or spreadsheet editing.

PowerPoint (.pptx) export is not currently available.

You can also share a conversation as a read-only snapshot link. The link is unauthenticated — anyone who has it can open the snapshot without signing in — so treat it as a public URL. A snapshot captures the messages as they existed when the link was created (later replies are not added), expires after 30 days, is limited to 10 active links per conversation, and can be revoked at any time. Citation titles, excerpts, and page numbers are included; document paths and file-download links are stripped before publishing. A signed-in colleague in your organization can also clone the snapshot into their own chat history.

Account & Team Management

Sevrel uses role-based access control to manage what each team member can do. Three roles are built in. Organizations that need finer-grained access can also have custom roles defined from the full permission set — those appear alongside the built-in roles when inviting a teammate or changing someone's role, but there is not yet a self-serve editor for creating them in the Company admin panel; contact support to configure them.

Creating a share link is available to any signed-in user, including viewers.

Organization administrators manage the team from the Company admin panel, not from Settings (Settings covers your own account only — profile, notifications, integrations, timeline, recently deleted, security, and privacy). From Company admin → Team an admin can invite teammates, change a member's role, and remove a member — removal revokes that member's sessions and unassigns them from the organization rather than deleting their account.