Notes & AI

Capture every client conversation with speech-to-text notes. AI extracts structured context, links similar past work, and turns your discovery into the next quote automatically.

Notes are the connective tissue of Servantium. They capture the raw context from every client interaction and feed the AI systems that power similar engagement matching, quote generation, and the institutional memory engine.

Creating notes

Notes can be attached to engagements, accounts, project plans, snippets, and other entities.

Open a record and navigate to the Notes tab, or click the explicitly labeled Notes button on specific items (like snippets, quote sections, project plan tasks, or resource assignments) to open the right-hand sliding popout. On desktop screens, these buttons display context labels (like “Quote Notes” or “Section Notes”) alongside their dynamic count badges.
Click Create to add a new note.
Enter the text in the free-form Note field.
Optionally use the speech-to-text button (microphone icon) to dictate your note.
Optionally click the Upload button to attach a file (up to 25MB) like a PDF, spreadsheet, or image.
Click Create to save.

Tip

Capture notes as soon as possible after a client conversation. The details you record become the fuel for Servantium’s AI features.

Managing notes and popout properties

The Notes tab displays all notes as a chronological list. Notes can also be accessed via the right-hand sliding popout across all workspaces, including engagement snippets, quote snippets, document variables, and the central Snippets library screen.

Notes Tab: Displays the discussion thread, semantic search, and file attachments. Notes count badges on buttons, snippet cards, and popouts subscribe to live database streams, updating in real time as soon as any note is saved without requiring a page refresh.
Properties Tab: Allows you to view and edit custom dynamic properties directly inside the popout window without leaving your workspace. If required fields are unpopulated, the Properties tab header displays a red warning dot alongside an inline alert in the form. Changes to quote section properties in the popout update the quote workspace in real time.
Semantic Note Search: A dedicated search bar at the top of the Notes tab allows you to query your notes conceptually. Because it uses vector embeddings, you don’t need exact keyword matches. Results display an AI similarity score (e.g., 85% match).
Include descendants: Toggle the hierarchy filter (tree icon) to view or search notes from the current record and all nested children (like project plan tasks) in a single unified view. When descendant notes are displayed in lists or search results, each note card renders an interactive target entity badge showing the record icon, collection name, and resolved record title. Tapping the badge navigates directly to the target record’s workspace.
Smart Note Routing: When you attach a note to a deeply nested record (like a specific project plan task), the system automatically links it to all parent records (like the overall project plan and the engagement). Additionally, if the note’s custom fields reference other records (like an account or a file), the system detects those references and ensures the note appears across all relevant workspaces automatically.
Editing & Pinning: Click into a note’s text to edit inline. Use the pin icon to keep important notes visible at the top of the list.
Attachments & Summaries: Attached files appear as interactive badges on the note card. Click the download icon to save the file. If AI has processed the document, click the sparkle icon to read a generated summary without leaving the page.

How notes power AI

Notes are the primary input for Servantium’s intelligence layer.

Note embeddings

When notes are saved, a backend trigger generates a vector embedding using Google’s Gemini model. This mathematical representation of the engagement’s context is stored for similarity searches.

AI context extraction

Click the AI button on an engagement’s Info tab to have AI read your notes and auto-fill template fields. This turns unstructured conversations into queryable data.