Capabilities reference.

Lorinks scopes extraction to three classifications: Contract, SOW, and Proposal. Other documents you upload (meeting attachments, briefs, decks) are kept as reference material but aren’t treated as sources of authoritative commitments. The point of the constraint is signal-to-noise — only the documents you’ve actually signed or formally exchanged become authoritative commitment data.

For each authoritative document, Lorinks extracts the commitments — who owes what, by when, on which clause. The platform is opinionated: it focuses on commitments, not on summaries, boilerplate, or stylistic content. Every extracted commitment is anchored to the exact clause in your document that produced it; if the platform can’t find that clause in the file, the item never reaches your screen.

Document commitments don’t live on the document — they surface in your Command Center, sitting alongside the items from your meetings as the reference layer for the engagement.

The Command Center opens with a meeting anchor strip at the top — the most recent meetings for this client, ordered by date, one click away. Beneath the strip sit three stat cards: Action Items, Questions, Risks — each counting the outstanding items in that category across the entire client relationship.

Below the stats, collapsible sections hold the actual rows. Items show owner, due date, severity, and a back-link to the meeting that produced them. Click any item for the full detail view: source quote, speaker, conversation, dispositions taken so far.

The since-last-meeting rail is a separate, collapsible section above the main item lists. It shows what changed since your last call with this client: new commitments made, items closed, scope discussions that didn’t exist a week ago. Once you’ve scanned them, a single mark-all-reviewed action clears the rail. It comes back at the next meeting.

Every item — actions, decisions, questions, risks — has the same disposition model: an unmarked default, plus two outcome states (Done and Invalidated). Dispositions live on the item, not on a global mode toggle. That means each meeting can change the state of an individual item without affecting the rest of the surface.

The portal’s engagement model is built around contribution where the work is owned. By default, a portal collaborator can comment and acknowledge on the action items, questions, and risks assigned to them — not on every item in the engagement. The principle is simple: everyone focuses on their own work. The client sees the same item ledger the consultant sees, but with comments and acknowledgments concentrated on the rows they personally own.

On their own items, a collaborator can mark On it in one click — an acknowledgment chip that becomes visible to every viewer in the portal in real time. The consultant sees the same acknowledgment as a notification inside Lorinks.

If a senior stakeholder needs to weigh in across the engagement rather than only on their assignments, the consultant can enable cross-item commentingfor that stakeholder. It’s a per-stakeholder option, not the default — focus stays the default, breadth is the exception.

Each stakeholder is invited at one of two access levels. Viewer is read-only — they can see items but not comment. Collaborator can comment on the items they own, following the engagement model above. Invitations are sent by email; the stakeholder clicks through, no password to set, no signup form to fill. If the invite expires, a single regeneration produces a new one.

Lorinks AI helps with drafts: pre-meeting briefs, opening lines for outbound update emails, visibility suggestions on newly-extracted items. But the consultant approves every word before any text reaches a stakeholder. Stakeholders never see “generated by AI” copy — they see what their consultant wrote, anchored to canonical items they can comment on or acknowledge.