Skip to main content
DoneThat

AI Adoption GuidePropertyRenew

Renewal Pipeline Workflow Agent

Agentic system tracks leases approaching renewal and break dates, triggers outreach sequences, routes negotiation tasks, and monitors progress to close across the portfolio.

Property processAcquireLeaseOccupyMaintainBillRenewVacateDispose

By Don, DoneThat’s AI coach · updated

What this agent does for a leasing coordinator

A renewal pipeline workflow agent watches the portfolio for leases that are approaching a renewal window or a tenant break date, then turns those dates into a work queue. It does not replace the coordinator. It flags which units need attention, drafts the next operational steps, and keeps a running view of whether each file is moving toward a signed outcome.

The coordinator still owns tenant conversations, commercial judgment, and the decision to send anything that has legal effect. The agent's job is speed of coverage: fewer units sitting unnoticed until a notice deadline has already passed, and fewer files stalled because nobody was assigned the next task.

Typical inputs are lease abstracts or the property management system's event calendar, unit and tenant identifiers, notice periods, option language (renewal, break, or both), and the current status of any live negotiation. Outputs are a dated worklist, draft outreach sequences, draft internal tasks for asset managers or surveyors, and a progress board that shows where each lease sits between identified and closed.

Related models feed this queue rather than replacing it. A Lease Break Probability Model can rank which break-date files deserve earlier human time. A Renewal Rent Optimization Model can attach a recommended rent band once a file is in play. Automated Renewal Heads of Terms can draft the commercial paper after terms are agreed. The pipeline agent is the traffic controller that starts and tracks that work.

How missing dates should fail closed

Event dates are the only reliable trigger. If the lease record has no renewal date, no break date, and no computed notice date the agent can defend, the agent should emit empty output for that lease: no sequence, no task, no reminder. Guessing a window from lease start plus a standard term is how portfolios send the wrong letter on the wrong day.

Coordinators should treat empty output as a data defect, not as nothing to do. The right follow-up is to send the file back to lease administration to complete the abstract, then re-run the agent. Partial dates are also a fail-closed case. A break option with a missing notice period cannot support a legally timed sequence. The agent can list the lease as blocked for incomplete event data on an internal exception report, but it should not invent a calendar.

When dates are present, the agent should compute working days backward from the contractual event, using the notice mechanics in the lease rather than a house rule. If the system cannot parse those mechanics, it should stop at a flagged exception instead of applying a default of 90 or 180 days.

Drafting outreach without sending notices

Once a valid window is open, the agent can draft a coordinator-facing outreach sequence: an internal brief, a suggested first tenant contact, a follow-up cadence, and a checklist of documents to have ready. Drafts stay in the work queue until a person reviews them. The agent must not auto-send tenant legal notices, break notices, statutory letters, or any message that could be read as exercising or acknowledging a formal right.

That constraint is operational, not cosmetic. A premature or mis-addressed notice can start a clock, waive a position, or create a dispute about whether the landlord acted in time. Email templates for a discussion of the upcoming lease are still drafts. Formal notice packs remain a human send, after the coordinator or counsel confirms parties, addresses, method of service, and the exact clause being relied on.

The same rule applies to routing. The agent can create a draft task for the person who usually handles that asset, attach the lease facts, and propose a due date aligned to the notice window. It should not mark the task complete, should not impersonate the coordinator in the tenant's inbox, and should not close the file because a draft was generated.

Routing negotiation work and watching the board

Negotiation is the coordinator's and asset manager's work. The agent's contribution is to make the file visible and to keep the next action dated. After first contact, typical routed tasks include: confirm tenant intention, collect comparable evidence, request a rent view from the optimization model, prepare heads of terms once commercial points are agreed, instruct solicitors, and diary the execution date.

The progress board should be boring and complete. Each lease in the active window needs a status, an owner, the next human action, and the next hard date (notice expiry, option expiry, or agreed response date). Stalled files are those where the next action is overdue and no new note has been logged. The agent can raise those internally. It should not chase the tenant automatically beyond the approved draft sequence, and it should not interpret silence as a break or a renewal.

Where break probability scores or rent bands are available, show them as inputs on the card, not as decisions. A high break-risk score is a reason to put a senior person on the file earlier. It is not a reason for the agent to serve a counter-notice or to pre-empt the tenant's choice.

What remains a human decision

Staff still decide whether to offer a renewal, at what rent and term, whether to accept a break, and when to involve lawyers. They still negotiate. The agent can pre-fill a task with facts (current rent, remaining term, option language, last correspondence date) so the meeting starts with a file rather than a hunt through the property management system.

Handoffs should be explicit. When heads of terms are ready, the coordinator opens the automated heads-of-terms draft, edits it, and only then circulates it. When a notice must go out, a named person sends it through the approved service method and files proof of service. The pipeline status then updates from that recorded action, not from the agent assuming the send occurred.

Auditability matters as much as speed. Keep a trail of which dates the agent used, which drafts it proposed, who approved or discarded them, and which messages actually left the building. That trail is what you need when a tenant disputes timing or when a colleague picks up a file mid-sequence.

Operating the queue day to day

Start the day with three lists: dates that entered a window overnight, files with overdue next actions, and exceptions with missing or unparseable event data. Work exceptions first if a notice period is already inside the risk zone, because no amount of drafted outreach helps a lease the system cannot date.

Then clear the new-window list: confirm the abstract, assign an owner, and release the first draft sequence for human edit. Finally, walk the stalled board with the people named on the cards. The agent can prepare that walkthrough. It cannot hold the commercial conversation.

If volumes spike at quarter-end, tighten the board rather than loosening send rules. Batch review of drafts is fine. Automatic service of tenant legal notices is not. Empty output on undated leases stays empty until lease admin fills the dates. That is the control that keeps a speed-focused pipeline from creating the next dispute.

Is this worth automating for you?

Whether this pays back depends on how much time it takes your team today. Most teams estimate that from memory, and the estimate is usually wrong in one direction or the other.

DoneThat reconstructs where the time actually went, with no timers to forget, so you can measure the baseline before committing to a project and check the gain afterward.

Measure the baseline first