Loading
Loading
A blueprint for one pipeline that catches every maintenance request, reads it, ranks its urgency, gets a manager's approval, dispatches the right vendor, keeps the tenant informed, and builds the owner report. People approve every dispatch and every closeout.
Solution blueprint: a system Callimachus Forge can architect for your operation. Not a case study, and not a claim of past deployment.
At a glance
Stack candidates
n8n · GoHighLevel · Twilio · Gmail · Airtable · Google Calendar · QuickBooks · Postgres
System run
A simulated run of the exact steps documented on this page, from first trigger to final handoff. Nothing here is live client activity.
Simulation · Loops
The problem
Maintenance requests arrive everywhere at once: a voicemail at midnight, a text to a coordinator's cell, an email with three photos, a portal ticket. Someone has to read each one, decide how urgent it is, and retype it into a tracker. A vague message about a water issue might be a drip or a burst pipe, and the difference is a ruined ceiling. Vendor dispatch is phone tag. Tenants call back for updates because they hear nothing. At month end, someone reconstructs what happened at each property from email threads to answer owner questions. The process runs on one coordinator's memory, and it breaks when volume spikes or that person takes a week off.
This blueprint fits a management company running scattered single family homes and small multifamily buildings for third party owners. A small office team handles intake, an answering service covers nights, and a roster of preferred vendors does the work. Each owner agreement sets a spend limit above which the owner must approve repairs. The maintenance coordinator sits between tenants, vendors, owners, and property managers, and nearly every step in that chain is a phone call, a text thread, or a spreadsheet row that only one person knows how to find.
The manual process
The automation opportunity
Every request, whatever channel it comes from, lands in one queue. Software reads the message and photos, identifies what kind of problem it is and how urgent, and pulls up the unit's history and the owner's spending rules. A manager sees a clean summary and approves the dispatch with one click. The right vendor gets the job with proposed times, the tenant gets automatic updates at every stage, and if a vendor goes quiet the job moves to the next one. When the work closes, the invoice and photos are already attached, and the owner's monthly report writes its own first draft. People still make every judgment call: the system does the collecting, sorting, chasing, and typing.
Architecture
Every step is typed: triggers, AI decision points, business logic, actions, data, and the human checkpoints that keep judgment where it belongs.
AI components
Human checkpoints
Tenant messages arrive by SMS, email, portal form, or after-hours call transcript. A webhook funnels every channel into one intake queue, so nothing lives only in a voicemail box.
A language model reads the message and photos, assigns a trade category and an urgency tier, and extracts unit, contact, entry permission, and pet notes into structured fields. Low confidence routes to human review.
The workflow pulls the unit's repair history, appliance and warranty data, and the owner's spend limit from the property database, and retrieves relevant lease clauses, attaching all of it to the request record.
Rules check hazard signals and the AI urgency tier. Water intrusion, gas, electrical hazards, or no heat in cold weather trigger the emergency path: an immediate phone alert to the on-call manager.
The property manager sees a one screen summary: issue, unit history, suggested vendor, and a cost band against the owner's limit. Nothing goes out to a vendor without this approval.
The matched vendor, chosen by trade, service zone, and current load, receives the work order by SMS and email with an accept or decline link and proposed visit windows pulled from scheduling.
If the vendor does not accept within the configured window, the workflow offers the job to the next vendor on the roster and notifies the manager, so requests never stall silently.
The tenant automatically gets plain language updates at each stage over their original channel: received, approved, scheduled with a date window, completed. AI drafts the wording inside approved templates.
The vendor submits photos and an invoice through a link. The manager spot checks the work and the amount before the work order closes, and mismatches go back to the vendor with a note.
The approved invoice posts to accounting against the correct property and owner ledger, and the work order record closes with final cost, dates, and attachments.
A monthly digest per property compiles requests, resolutions, and spend from work order data. AI drafts the narrative summary, a manager reviews it, and the report goes to the owner by email.
Engineering notes
Benefits are qualitative by design. We do not publish numbers we cannot verify.
The free audit maps your actual tools and process against this architecture, and tells you honestly what is worth building.