Build Your After-Hours AI Answering Service
The best providers are either large call centers with high monthly fees or a custom-built AI system. A custom system answers tenant calls instantly with full context from your property management software.
We built a system for a property manager with 25 buildings and 400 residential units. They were paying a call center to field after-hours calls, but urgent requests were still missed. We built an AI agent that triages calls in 90 seconds and automatically creates work orders in AppFolio. The system went live in 3 weeks.
The scope is driven by the number of unique call types and integration points with your Property Management Software (PMS). A system that only handles maintenance requests is simpler than one that triages ten different issue types and dispatches vendors automatically. An engagement is scoped based on these specific call flows.
What Problem Does This Solve?
Most property managers start with a traditional answering service. This creates an 8-12 hour delay for non-urgent issues because the operator simply takes a message for staff to process the next morning. For emergencies, they page an on-call tech, but they lack access to your PMS to verify tenant details, leading to critical errors.
A tenant calls at 10 PM about a burst pipe. The call center operator, reading from a generic script, takes down the name and unit number. They mishear the unit as '3B' instead of '3D' and page the maintenance tech. The tech wastes 20 minutes at the wrong apartment while thousands of dollars in water damage occurs in the correct one. The root cause is a human data entry error from a non-integrated service.
Basic IVR systems from providers like RingCentral are no better. They force callers into rigid menus ('press 1 for maintenance') and cannot distinguish a slowly dripping faucet from a flooded apartment. Both calls end up in the same voicemail box, and every issue is treated with the same low priority until a human listens to the messages hours later.
How Does It Work?
We connect a Twilio phone number to an AWS Lambda function running a FastAPI application. When a call comes in, the system authenticates the caller's number against your tenant roster via an API call to your PMS, like Buildium or Yardi. This takes less than 200ms and means the system knows who is calling and where they live before they even speak.
The live audio stream is processed by the Claude API for transcription and intent recognition. We configure the agent to classify calls into specific categories: emergency maintenance, routine repair, lockout, or noise complaint. It distinguishes a 'pipe burst' from a 'dripping faucet' based on keywords and sentiment, assigning an urgency score from 1 to 5.
Based on the urgency score, the system takes immediate action. A score of 5 triggers an API call to PagerDuty to alert the on-call technician. A score of 1-4 creates a detailed work order directly in your PMS with the full transcript attached. For lockouts, it verifies the tenant's identity with a security question and sends the lockbox code via SMS. A full call is processed and resolved in under 90 seconds.
Every call, transcript, and action is logged to a Supabase database for a complete audit trail. We use structlog and AWS CloudWatch to monitor the system. If an API call to your PMS fails more than three times, an alert is sent to a designated Slack channel for immediate review. Hosting costs on AWS are typically under $40/month.
What Are the Key Benefits?
Resolve Tenant Issues in 90 Seconds
From incoming call to a work order created in your PMS in under two minutes. No more next-day data entry or missed messages.
Pay For Usage, Not Per-Agent Seats
Your monthly bill is based on Twilio and AWS usage, not a fixed retainer. A single fixed-price build replaces recurring call center fees.
You Own the System and Phone Number
We deliver the full Python source code to your GitHub repository. You are not locked into our service or any proprietary platform.
Automatic Escalation for True Emergencies
The system uses PagerDuty to immediately contact your on-call staff for P1 issues like fires or floods, including a full call transcript.
Writes Directly to AppFolio or Yardi
We build direct API connections to your specific PMS. Work orders appear instantly with all necessary details, formatted correctly.
What Does the Process Look Like?
System Scoping (Week 1)
You provide API access to your PMS and walk us through your current after-hours protocol. We deliver a System Design Document detailing call flows and integration points.
Core Logic and API Build (Week 2)
We build the FastAPI application and integrate the Claude API for language understanding. You receive a private phone number to test the core call classification logic.
PMS Integration and Deployment (Week 3)
We connect the system to your property management software and deploy it to your AWS account. You receive a staging environment for end-to-end testing.
Launch and Handoff (Week 4)
We switch the system to live. After a two-week monitoring period, we deliver the final source code, documentation, and a runbook for your team.
Frequently Asked Questions
- What drives the cost and timeline?
- The main factors are the number of integrations and the complexity of your call logic. A system that only handles maintenance requests for one PMS is a 3-week build. A system that handles maintenance, lockouts, and leasing inquiries and connects to both AppFolio and a separate accounting tool might take 4-5 weeks.
- What happens if the AI can't understand a caller?
- If the system cannot determine the caller's intent after two prompts, it automatically transfers the call to a human on-call number. It also logs the transcript fragment for review, allowing us to improve the recognition model over time. This fail-safe ensures no tenant request is ever dropped.
- How is this better than a service like PATLive?
- PATLive and other virtual receptionists are humans following a script. They cannot access your PMS to verify tenant identity or create work orders directly. They take a message, which creates a delay. Our system is integrated, resolving issues in real-time and eliminating next-day manual data entry.
- Does it work for tenants who speak Spanish?
- Yes. We use the Claude API, which supports multiple languages. We can configure the system to detect the language being spoken and respond appropriately. The transcript and generated work order can be automatically translated back into English for your staff, with the original Spanish transcript attached for reference.
- What if we switch our property management software next year?
- Because you own the source code, you can hire any developer to update the integration. The core logic for understanding calls is separate from the PMS connection. We typically estimate that swapping one PMS API for another is a 3-5 day project, not a complete rebuild.
- Can the system also respond to text messages?
- Yes. The same core logic can be exposed through a different Twilio endpoint for SMS. A tenant could text 'my sink is leaking in 4B' and the system would create the same work order as if they had called. Adding an SMS channel is a common extension and typically adds 2-3 days to the initial build.
Related Solutions
Ready to Automate Your Small Business Operations?
Book a call to discuss how we can implement ai automation for your small business business.
Book a Call