Skip to content

AI agentsAI property management

AI property management, with a person in charge.

WiseBlock's agents do the reading, sorting, matching and chasing that fill a block manager's week — emails, bank statements, arrears, repair photos, contractor shortlists. Each one is switched on and configured by you, and the decisions that commit money or speak for your firm stay with a person.

The AI Agents settings page in WiseBlock: a count of agents active, then a card per agent — Email Triage and Bank Specialist marked active with a Deactivate button, Compliance Monitor marked coming soon — each listing its capabilities.

Sheet 01The settings page

Every agent is off until you turn it on.

Agents live under Settings, one card each, with the list of what it can do. An organisation administrator activates the ones they want, configures them, and can deactivate any of them at any time. Every request an agent makes is logged with the tokens it used and what it cost, so usage is never a surprise.

  1. Active, and since when.

    An active agent shows the date it was switched on, and its capabilities in plain words.

  2. Coming soon means not yet.

    Agents still being built are labelled as such and cannot be activated. We list them so you know what is next, not to sell them.

  3. Deactivate or configure.

    Each live agent can be switched off or opened to change how it behaves — thresholds, schedules, what it drafts.

Sheet 02The agents

Ten agents, and what each leaves to you.

The full list, including the three that are still being built. The third column is the point: what the agent hands to a person rather than doing itself.

WiseBlock AI agents, what they do and what a person decides
AgentWhat it doesWhat a person decidesStatus
Email TriageClassifies incoming building email (maintenance, compliance, finance, resident query, spam), sets a priority, labels it and flags urgent items.Whether to send the draft replies it suggests.Available
Bank SpecialistReads PDF and CSV bank statements, extracts every line, creates ledger entries and matches them for reconciliation.Lines it marks needs review — usually your accountant.Available
Money CollectorCalls residents whose balance is over your threshold with a scripted reminder, within the days and hours you set, by Twilio voice.The threshold, schedule, call limits and wording. Residents can opt out of automated calls.Available
Maintenance SurveyorReads photos of a fault and drafts the title, description, category, urgency and trades needed.Everything: it fills an editable form, and nothing is saved until someone checks it.Available
Contractor MatcherScores eligible contractors on trade, locality, responsiveness, quality, price and workload, from your own records, with the reasons.Who is asked to quote, and who wins the work.Available
AI ReceptionistAnswers residents by chat and voice in the resident portal, around the clock, using their own account.Anything it can't settle goes to your team as a contact or repair request.Available
Platform AI AssistantAnswers your team's questions about the portfolio from live data — buildings, residents, balances, overdue checks.What to do with the answer. It won't send data outside, and points to Reports for exports.Available
Compliance MonitorWill track fire safety certificate expiry and alert before deadlines.—Coming soon
Financial AnalystWill flag unusual spending and compare budgets across buildings.—Coming soon
Board ModeratorWill moderate building message boards and summarise long threads.—Coming soon

Sheet 03Agents in a chain

Two agents and a property manager fix a leak.

The Maintenance Surveyor reads the resident's photo and drafts the job; the Contractor Matcher finds who to ask. The price is checked against the budget and the approval limit — and a person awards it.

  1. Reported
  2. Analysed
  3. Quotes requested
  4. Quoted
  5. Approved
  6. Fixed

An illustrative example: the building, residents, contractors and figures are fictional.

Sheet 04Where people decide

Agents draft and sort. People commit.

Three places in WiseBlock are a person's decision by design, whatever the agents are doing.

Awarding work. A quotation over the building's approval threshold can't be awarded until someone approves it — and the threshold is tested against everything already let on the same job, so a large repair can't be split into small ones to slip under it. Section 20 consultation remains your process; WiseBlock keeps the figures honest.

Approving invoices. Contractor bills are checked against the work order, then approved by a person before they reach the ledger.

Needs review. When an agent isn't sure — a bank line that fits two flats, a photo that can't show what's behind the wall — it says so and hands it over, rather than guessing.

Sheet 05Configuration

Set each agent up the way your office works.

Configuration is per organisation, from the agent's own settings screen. A few examples of what can be changed:

  • Money Collector

    The balance above which a resident is called, the weekdays, weekend days and hours calls may be made, how many attempts, and the message — with the resident's name, amount and payment reference filled in.

  • Maintenance Surveyor

    Which fields it drafts (category, urgency, trades), how many photographs it reads, from one to eight, and standing instructions added to every request.

  • Contractor Matcher

    How much each signal weighs — or switch one off — the shortlist size, the minimum score, and whether unverified firms may be suggested.

  • Usage

    Every request is logged against the agent with its tokens and cost, so you can see what each one is doing and what it costs to run.

Questions

AI agents, answered.

Can an agent spend the building's money?

No. Agents draft, sort, match and remind. Awarding work over the approval threshold needs a named person's approval, and contractor invoices are approved by a person before they reach the ledger.

Which AI models do the agents use?

Different agents use different models: Anthropic's Claude for the platform assistant, email and bank statement work, OpenAI's vision model for reading repair photographs, and VAPI for the receptionist's voice. Usage is logged per request.

Do we have to use all of them?

No. Every agent is off until an administrator activates it, and any of them can be deactivated at any time.

Will the Money Collector call residents who've asked not to be called?

No. Residents can opt out of automated calls from their portal profile, and calls only happen within the days, hours and limits you set.

Are the coming-soon agents available?

Not yet. Compliance Monitor, Financial Analyst and Board Moderator are listed so you can see what's planned; they can't be switched on until they are built.

Next step

Put the agents to work on your own blocks.

A 30-minute demo on a portfolio like yours: fire safety, repairs, residents and service charges, with the agents switched on.