AI agents

Meet the 10 AI agents behind the busy work

Each agent has one job. Each one shows its reasoning. And you approve what matters, so the personal touch with your guests stays yours.

30 days free on up to 10 units · no card · FREE one-on-one onboarding with a real person, or simply through our onboarding AI

Your team at a glance

Ten agents, one job each, working 24/7

Repairs, vendors, turns, arrivals, reviews, calendars, invoices, leases and your morning briefing. You approve what matters.

Watch one repair

One broken AC, six agents, no phone tag

This is a sample story, not a customer result. It shows how the agents hand work to each other.

  1. Maintenance Coordinator Agent

    A guest writes: “The AC isn’t cooling.”

  2. Vendor Dispatcher Agent

    Picks the best vendor on real history.

  3. Vendor Chase Agent

    No reply after two hours.

  4. Readiness Agent

    Next guest arrives Friday.

  5. Review-Risk Agent

    Checkout day.

  6. Concierge Agent

    Next morning.

Rule 1

AI only where words are messy

Facts, dates, money and rankings are plain math from your own records. AI is used only to read free text, or to write a sentence, never to make up a number.

Rule 2

Every decision has a reason

Each step an agent takes is written to your activity log with why it did it. You can read the whole trail for any repair, turn or alert.

Rule 3

You set the limits

Approval limits, quiet hours, and “always ask me first” rules by unit or trade. For vendor, supplier and invoice messages you choose how much goes out on its own.

What each agent does

Eight of the ten work on every plan, including Lite. The Lease Agent and the Invoice and Parts Chase Agent come with Starter and up.

Maintenance Coordinator Agent

Reads a repair request in plain words and works out the trade, how urgent it is, and whether it is the owner’s or the tenant’s job. If a guest is in the unit, it treats the same fault as more urgent.

For example

“AC isn’t cooling, it’s 95 degrees outside” → HVAC, emergency, guest inside.

  • You stay in control: Every call is logged with its reason, and you can change it.
  • How it decides: AI reads the words; the urgency and ownership rules are plain logic.

Every plan, including Lite

Vendor Dispatcher Agent

Ranks your vendors on their real history (completion rate, rating, cost) and sends the job. Under your approval limit it dispatches; over it, it waits for your OK.

For example

“Cool Air HVAC: 97% completion, 4.7 rating, under your $500 limit” → dispatched, reason saved.

  • You stay in control: You set the approval limit, and rules like “always ask me first” for a unit.
  • How it decides: Ranking is math on your own job history, not a guess.

Every plan, including Lite

Vendor Chase Agent

Nobody shows up and nobody notices is where bad reviews start. When a vendor goes quiet it nudges them, tries again, then hands the job to you before a tenant has to call.

For example

Day 1 no reply → re-email. Still nothing → firmer nudge, you are copied → it hands the job to you.

  • You stay in control: You choose draft-only, routine messages only, or fully automatic for each vendor.
  • How it decides: A fixed three-step ladder. No AI improvising.

Every plan, including Lite

Turn Agent

When a short-term guest checks out it opens the cleaning turn on its own, sets the next check-in as the deadline, and matches a cleaner from your vendors.

For example

Guest out Thursday 11 am, next guest in Thursday 4 pm → same-day turn, hard deadline.

  • You stay in control: You can reassign any turn, and the deadline is shown on your calendar.
  • How it decides: Calendar facts and rules, no guessing.

Every plan, including Lite

Readiness Agent

Checks every arrival in the next two days against open repairs and unfinished cleaning, and warns you before the guest finds out. It can also email an owner a one-page “is the unit ready” report.

For example

Arrival Friday, dishwasher repair still open, cleaning not started → alert Wednesday.

  • You stay in control: One alert per arrival, not one every hour. The report is sent only when you ask.
  • How it decides: Calendar, repairs and turns checked together. All rules.

Every plan, including Lite

Review-Risk Agent

At checkout it looks for stays where something went wrong, like an AC request that stayed open for 30 hours, and drafts a short note you can send while it still matters.

For example

2-night stay, AC open 30+ hours → flagged, apology drafted for you.

  • You stay in control: It drafts. You decide whether to send, and what to say.
  • How it decides: Severity is a rule. AI only drafts the note.

Every plan, including Lite

Calendar Guard Agent

Reads your Airbnb, Vrbo and Booking.com calendars, catches two bookings on one night, and warns you when a calendar link stops updating, which is how double bookings usually start.

For example

Two reservations share Oct 24 → you are emailed the same hour, with both stays.

  • You stay in control: It is read-only. It tells you; it never changes your channels.
  • How it decides: Dates compared by rules. No AI.

Every plan, including Lite

Invoice and Parts Chase Agent

Reminds a customer about an overdue invoice and a supplier about a part that is holding up a turn, on a polite three-step ladder, with you copied on the firm one.

For example

Part due Monday, nothing shipped Tuesday → supplier nudged, the turn it blocks is flagged.

  • You stay in control: Per invoice or supplier you pick draft-only to automatic. It never takes a payment.
  • How it decides: A fixed ladder. No AI improvising.

Starter and up

Lease Agent

Starts leases on their start date, ends them on their end date, and flags renewals where the notice window is closing, so a lease behaves like a system and not a spreadsheet.

For example

Offer sent, 90-day notice window closing → flagged on your morning list.

  • You stay in control: Renewals and notices stay yours. It only flags.
  • How it decides: Dates and rules only. No AI.

Starter and up

Concierge Agent

Every morning it tells you the few things that need you today, and answers plain-language questions across your whole portfolio, long-term and short-term together.

For example

“Which units earned the least this year, and what did repairs cost on them?” → a ranked answer from your own data.

  • You stay in control: It answers from your data. It cannot send, pay or change anything on its own.
  • How it decides: Every number is counted from your records. AI only writes the sentence.

Every plan, including Lite

Also in the box

They work with these, too

  • Guest emails, drafted for you. Check-in and thank-you emails are written from your own instructions. Nothing goes to a guest until you read it, edit it and approve it.
  • Rates that show their reasoning. Nightly suggestions list why each night is priced the way it is.
  • Damage claim packs. It builds the pack and counts down the deadline. You file it.

What they never do

Lines the agents don't cross

  • ✓ Send a guest a message you haven't approved
  • ✓ Collect, hold or move money. Invoices and parts are chased, never paid through TraxKey
  • ✓ File a damage claim for you
  • ✓ Change your Airbnb, Vrbo or Booking.com calendars. They are read-only
  • ✓ Spend over your approval limit without asking

Put the team to work on your portfolio

Thinking about switching? We make it easy: FREE one-on-one onboarding with a real person, or simply through our onboarding AI, which imports your properties, units, leases and vendors.

Start 30 days free

Up to 10 units, no card. The glowing orbs are the agents, not people.