Blog / Case study
An office broker uses AI to find companies that may need space before they start looking
We built an AI research and CRM workflow for an office tenant representation broker. It monitors public records for signs that a company may need to expand, relocate, or reduce its office space.
Each prioritized prospect arrives with a reason to call, supporting sources, an estimated lease timeline, a likely decision-maker, and a prepared outreach draft. The broker receives a weekly shortlist of up to 25 companies and alerts when a prospect reaches the highest priority tier.
- Daily
- Monitor company activity and update scores
- Weekly
- Deliver up to 25 prioritized companies
- When a prospect becomes Hot
- Create a call task and prepare an email draft
The problem
A broker’s best opportunity often starts with a change inside a company: a funding round, a hiring push, a relocation, or a reduction in staff. Each event can affect how much office space the company needs.
Those signals appear across public filings, job boards, construction permits, and business announcements. Finding one is only the beginning. The broker still has to connect it to the right company, establish where that company operates, assess its likely timing, find a decision-maker, and prepare a reason to reach out.
For a small brokerage, that research competes with client meetings, tours, negotiations, and the rest of the work required to close a lease.
The goal was to give the broker a researched reason to contact a company, with the supporting information already in the CRM.
What we built
The workflow collects public records and looks for events associated with changing office requirements. AI extracts relevant details from those records, helps match them to companies, and prepares a brief for prioritized prospects.
The system combines that research with a scoring rubric and an estimated lease window. It then updates the CRM and creates the appropriate follow-up.
| Stage | What the system does | What the broker receives |
|---|---|---|
| Discover | Monitors filings, hiring, permits, registrations, and announcements | A dated record of relevant company activity |
| Connect | Matches related records to a company | A single profile with supporting sources |
| Qualify | Scores signals and estimates lease timing | A ranked prospect with an explanation |
| Prepare | Finds a likely contact and uses AI to draft outreach | Contact details, a research brief, and a first email |
| Follow up | Updates the CRM and creates tasks | A shortlist and reminders for the right opportunities |
The broker reviews the research and sends the outreach. Every AI brief points back to its underlying sources.
Finding the companies worth a call
The system looks for events that can change a company’s office requirements.
| Signal | What it can indicate |
|---|---|
| Funding and hiring growth | A company may need more space |
| Relocation or expansion announcements | A company may be planning a move or entering a market |
| Layoffs and closures | A company may need to reduce space or consider a sublease |
| New business registrations | A company may be establishing a new location |
| Office permits and fit-outs | A company has invested in a location, providing an anchor for future lease research |
| Public lease disclosures | A filing may contain an actual lease term or expiration date |
A construction record needs context. A recent fit-out may mean the company has already chosen its space, so the system lowers its immediate priority. The same record can become useful later as evidence of when the company moved in.
Connecting the evidence
A company can appear under different names across a permit, a job posting, and a public filing. The workflow matches records using identifiers such as company domains and filing numbers, then compares names and addresses where necessary.
For each company, the system gathers available office information and identifies a likely decision-maker. It targets a CEO or founder at smaller companies, or a COO or CFO at mid-sized companies. Contact records show their source and whether an email has been verified.
Where square footage is unavailable, the system can estimate it from headcount and labels the figure as an estimate. It also avoids treating a registered-agent address as proof of the company’s operating office.
Estimating when to reach out
A company’s lease expiration is often private. The workflow uses available evidence to establish a possible window and records how confident it is.
| Evidence | How the system uses it | Confidence |
|---|---|---|
| Broker-confirmed date | Uses the date provided by the broker | High |
| Public lease disclosure | Uses the stated term or expiration | High |
| Fit-out or move-in record | Estimates timing using an assumed lease term | Medium |
| First observed activity at an address | Estimates timing from the earliest available record | Low |
The initial lease-term assumptions vary by office size: three to five years for spaces under 10,000 square feet, five to seven years for 10,000 to 50,000 square feet, and seven to ten years for larger spaces.
The estimate helps the broker decide when to investigate. Once the broker confirms an expiration date, that date replaces the estimate.
The scoring rubric
The system scores companies from 0 to 100 using the timing and strength of their signals. Older signals lose weight over time.
| Signal | Points | Weight falls by half after |
|---|---|---|
| Estimated lease expiration in 9 to 24 months | +30 | Recalculated daily |
| Announced relocation or expansion | +25 | 90 days |
| Funding raise of at least $5 million with a local address | +20 | 120 days |
| Layoff or closure notice in the market | +20 | 90 days |
| Local openings increase 50% in 90 days, or reach at least 10 roles | +15 | 60 days |
| Out-of-state company registers in the market | +15 | 120 days |
| Estimated space falls below 150 square feet per person | +15 | Recalculated daily |
| New H-1B workers at a local worksite | +10 | 180 days |
| New space fit-out within the past six months | −30 | Revisit in three years |
| Estimated company size under 15 employees | −20 | Configurable |
The score ranks opportunities for review. It does not represent a measured probability that a company will move.
| Priority | Score | Action |
|---|---|---|
| Hot | 70 or higher | Alert the broker, create a call task, and save an email draft |
| Warm | 50 to 69 | Include the company in the weekly shortlist |
| Watch | Below 50 | Keep the company in the CRM and continue monitoring |
From research to outreach
For each Warm or Hot company, AI writes a short brief covering what happened, what it may mean for the company’s office needs, and a possible opening angle.
A broker might see that a company raised funding and increased hiring in the same market. The CRM record brings those signals together with the company’s location, estimated timing, and source links.
For Hot prospects, AI also prepares a first email based on a specific public signal. The broker can review the evidence, edit the draft, and decide whether to send it.
The workflow creates a call task for new Hot prospects and a re-engagement task when a company enters its estimated decision window.
The impact
The broker receives a recurring list of researched prospects, with the information needed to decide whom to contact.
| Activity | Before | With the AI workflow |
|---|---|---|
| Find prospects | Search public sources individually | Monitor sources on a schedule |
| Understand a company | Connect filings, hiring, and location records manually | Review a company profile with dated sources |
| Decide whom to contact | Assess each opportunity from scratch | Use scores, timing estimates, and research briefs |
| Prepare outreach | Find a contact and write the first email | Review a likely decision-maker and AI draft |
| Manage follow-up | Enter research and create reminders manually | Receive CRM updates, alerts, and tasks |
The case study does not yet have measured totals for hours saved, meetings booked, or revenue generated.
Measuring commercial results
The workflow tracks what happens after a prospect enters the CRM: whether the broker holds a meeting, signs a representation agreement, and closes a lease.
That gives the broker a way to see which signals produce business. Hiring growth might generate meetings, while an approaching lease expiration might lead to more representation agreements. Over time, those results can guide which companies receive attention first.
| Measure | Business value |
|---|---|
| Signals that lead to meetings | Identify events that prompt useful conversations |
| Meetings that lead to representation agreements | Assess whether the prospects have a need the broker can serve |
| Agreements that lead to signed leases | See how sourced opportunities progress through the pipeline |
| Commission from sourced prospects | Compare the system’s cost with the business it generates |
| Poor fits and incorrect matches | Reduce repeat mistakes and unnecessary review |
The broker can also label each lead as useful, already known, unsuitable, or incorrectly matched. That feedback supports changes to the scoring weights and helps keep poor fits out of future outreach.