Spector Gadget is the planned iOS + Android app for Spector — the tenant-prospecting engine that works a commercial vacancy every business day until it fills, then ends its own subscription in celebration. This paper describes what the broker holds in their hand: the mission dashboard, the daily lead feed, one-tap approve-and-send outreach, and the Vacancy Filled Receipt.
The problem: a vacant commercial suite costs its landlord rent every day, and the listing broker's prospecting time is the scarcest input in filling it. Most vacancies are marketed passively — a flyer, a LoopNet entry, and hope.
The engine (live today): Spector is an AI agent that actively prospects a listed vacancy every business day: rotating through news, franchise-development wires, and ground-truth sources; deduplicating and qualifying named candidate tenants with a reason and a public source for each; and drafting personalized outreach that the broker approves and sends. It runs until the vacancy fills — then the subscription ends itself and mints a commemorative Vacancy Filled Receipt.
This paper (the app): today that engine speaks email. Spector Gadget wraps it in a phone-native client — because brokers live on their phones, between showings — turning the daily loop into a lead feed, the approval into one tap, and a prospect's reply into a push notification the broker can act on in the parking lot.
Every listing broker knows this math. Few have a tool aimed squarely at it.
A 2,700 SF retail suite at ~$20/SF NNN is roughly \$4,500/month of rent not being collected — every month it sits. (Illustrative; your suite's number is on your own flyer.)
Brokers juggle many listings; proactive tenant-hunting for any one of them happens in stolen minutes. Signals — a chain signing a franchise deal, a food truck telling a reporter it wants a brick-and-mortar — scroll past unseen.
The commission on a filled lease is thousands of dollars. A prospecting tool should cost a rounding error of that and leave the relationship — and the win — with the broker. That ratio is Spector's design constraint, not a marketing line.
What the app feels like to a working broker, from listing to lease. Everything below maps to engine behavior that exists today (see §3) — the app changes the surface, not the substance.
Snap or paste the flyer, confirm address and suites, note any deal facts (NNN vs gross, restrictions, "never pitch competitors of my anchor"). Subscription starts: \$36.90/mo for this property. The meter is mission-shaped — it stops the day the space fills.
A full prospecting report: trade-area facts with sources, a tenant-fit thesis, 15–25 named candidate tenants ranked by likelihood-to-transact — each with a one-line reason, a public source link, and a contact path. Readable in the app, forwardable as a branded PDF.
Spector rotates sources daily — business journals, franchise wires, ground truth. New qualified leads appear in the feed with their reason and source. Quiet days say so honestly: "3 sources checked, no new candidates, list still fresh."
Tier-1 leads come with a personalized outreach draft. The broker reads it, edits if they like, and taps Approve & Send — the note goes out from the broker's own identity. Spector proposes; the human sends. Always.
New leads this week, the re-ranked call-first five, drafts awaiting a tap, movement (replies, retirements with comp intel), and what Spector needs from you. One screen, no filler.
A push notification with the reply summarized, a suggested response drafted, and the prospect's context card — actionable from a parking lot between showings. Speed-to-response is where deals are won.
Tap Mark as FILLED. Billing stops that day. Spector mints the Vacancy Filled Receipt — a branded certificate: engine start, lease date, days-to-filled, the journey. The only subscription that celebrates its own cancellation.
This is the honest mechanics — not app fiction. The engine operates now over email; the app is a thin client over exactly this loop.
Every pass starts by reading what changed: broker instructions, new deal facts, prospect replies, a suite leasing. A leased suite retires every lead sized to it — the thesis re-ranks before any new prospecting.
Different source families each weekday — local business press, food & retail media, franchise-development wires and state feeds, brand location-page truth checks, and a Friday ground-truth sweep of the trade area. Everything gets covered weekly without hammering anything daily; every scan is logged.
Every surfaced business is canonicalized and checked against the property's lead ledger. Known lead → signal update, not a duplicate. New candidate → the full rubric: expansion-intent tier, footprint vs. available suites, co-tenancy conflicts, economics sanity, recency, callability. Rejections are kept, with reasons, so nothing is re-researched.
Qualified tier-1/2 leads get a personalized draft: a hook proving homework was done, one sentence of numbers that matter to their model, an honest opt-out. The broker approves and it goes out under the broker's identity. Spector cannot send prospect mail — structurally, not just by policy.
| Day | Source family | Why it matters |
|---|---|---|
| Mon | Local business press (business journals, CRE trades) | Lease signings, expansions, relocations announced over the weekend |
| Tue | Food & retail media, "what's opening" trackers | Operators publicly stating they want their next location |
| Wed | Franchise wires + state-level development feeds | "Signed a multi-unit agreement in this state" = a tenant actively hunting sites |
| Thu | Brand-truth day: franchise-dev & location pages of watched leads | Detect new "coming soon" pins and footprint-spec changes at the source |
| Fri | Ground truth: maps/review sweep of the trade area + competing listings | New businesses, closures (= space users in motion), rate comps |
| Daily | News catch-all query on the corridor | Two minutes of cheap insurance against missing anything |
| Monthly | Civic: planning agendas, urban-renewal minutes, grant awards | Slow-moving, high-signal — who's investing in this corridor |
Every pass opens by incrementing the day counter and asking one question: what gets this vacancy filled faster today? The loop's success condition is its own termination. When the broker confirms the lease, billing stops, the receipt is minted, the workspace is archived intact for the client, and days-to-filled joins Spector's public scoreboard — which is only ever updated with real missions, including the ones that end without a fill.
Six core screens. Lead names below are drawn from Spector's real free sample report (Shops on Wadsworth, Arvada CO); scoreboard numbers are illustrative placeholders — the live app shows only real missions. Swipe →
🔥 Chef Tom replied — asking about TI and patio. Suggested response ready.
v1 ships the engine's existing surfaces as native screens. v2 adds the things only an app can do well. Nothing here changes the guardrails.
| Feature | v1 (thin client) | v2 |
|---|---|---|
| Missions dashboard | All engine properties, day counters, open suites, queue depth | Portfolio view for teams; per-suite mission tracking |
| Lead feed | Living ranked list; every lead = reason + tappable source; retired leads as comp intel | Filters by suite/tier/category; swipe to decline with reason (feeds the ledger) |
| Approve → send | One tap opens a pre-filled draft in the broker's own mail client (send is theirs) | In-app send-on-behalf after explicit opt-in: reply-to broker, CAN-SPAM-clean, still human-approved per message |
| Inbox triage | Prospect replies summarized + suggested response + context card | Full thread view; call-notes capture ("log what happened on the call") |
| Push notifications | Prospect replied · hot new tier-1 lead · Monday digest ready | Quiet-hours, per-mission notification tuning |
| Scoreboard | Days-on-engine per mission; public avg days-to-filled (real missions only) | Filled Receipt gallery; shareable receipt cards |
| Onboarding | Flyer photo/PDF → extracted facts → broker confirms → engine starts | MLS/listing-link import; brochure auto-parse improvements |
| Team seats | — | Broker + assistant + landlord read-only view of the mission |
| Billing | Stripe subscription per property; FILLED tap stops the meter | In-app purchase parity where stores require it; receipts archive |
The economics are part of the product design, stated plainly.
\$36.90 per month, per property — until it fills. One property, up to ~4 suites, one trade area. The subscription is mission-shaped: it is supposed to end, and its natural ending is a celebration, not churn. When the broker marks the vacancy FILLED, billing stops that day and the record converts into the commemorative receipt.
96.3% of the value stays with the broker. A \$36.90/mo tool that helps land a lease worth a five-figure commission keeps essentially all of the win — and the relationship — with the human who made the call. Spector's split: 0.33 → founder, 3.36 → company/compute, 96.3 → the client's side of the table.
Propose, never send. Every prospect-facing message is drafted by Spector and sent by a human, under the human's identity. In v1 this is structural — Spector has no path to send prospect mail at all.
Public data + honest outreach only. No scraping behind logins, no fake identities, no spam. Every lead carries a public source link; unverifiable numbers are labeled as assumptions, never asserted. Sources that block bots are noted, not circumvented dishonestly.
The guarantee. If the onboarding report doesn't genuinely help, the month is refunded. The scoreboard reports all missions — including the ones that end unfilled.
A white paper should say what's real. Here's exactly what is.
What Spector is. Spector is a persistent AI agent running on the Taurus multi-agent platform — durable infrastructure with its own workspace, memory, and scheduled runs. The prospecting engine (daily source rotation, lead ledger with dedup, qualification rubric, outreach queue, digest, filled-flow) is specified and operating today, delivering over email and shared documents. Real reports for real Colorado listings have shipped.
What the app is. Spector Gadget is a thin client: the phone renders state the agent already maintains (missions, ledgers, queues, digests) and sends back the broker's decisions (approve, decline, facts, FILLED). No intelligence moves into the phone; the app is a window, which is why the email-first engine and the app can honestly coexist — same loop, two surfaces.
What is live vs. roadmap. Live the engine, the reports, the pricing covenant, the storefront at spectoragent.com. Roadmap everything in Section 4–5: the native apps, push notifications, in-app approve, team seats. Scoreboard numbers shown in mockups are illustrative; the public scoreboard launches empty and fills only with real missions. We will not claim an average days-to-filled we haven't earned.
Data handling. Broker-provided deal facts stay in the mission workspace, delivered to the broker; they are not resold or shared across clients. Lead research uses public sources only.
Onboarding reports, daily loop, Monday digest, mailto one-click approvals, filled-flow + receipt pipeline. Free full sample on the storefront.
The six screens of Section 4 as an installable web app — fastest honest path to "in your pocket," push included, no store gatekeeping.
Store presence, native push, flyer-camera onboarding, in-app approve→send (opt-in), team seats, receipt gallery.
Each stage ships only when the previous one is earning real missions. The metric that gates everything: are vacancies actually filling faster?