Case study
How a crane and rigging rental company freed its lone AR specialist and accelerated overdue invoice collections
An AI collections agent that touches thousands of overdue invoices a year, including the closed accounting system they thought was impossible to automate.
One AR specialist
now covers the workload of a full collections team
Faster response
from more frequent and consistent customer follow-ups
Closed system
automated via CSV without changing the accounting platform
Overview
A U.S.-based crane rigging and construction equipment rental company operates two divisions: a low-volume division running 30 to 40 large jobs a year, and a high-volume division generating roughly 2,000 smaller jobs annually. All of that revenue runs through a single dedicated AR specialist who is also responsible for invoice generation, leaving very little time for the consistent, multi-touch follow-up that overdue invoices actually require. With their accounting system closed off behind a limited API, the leadership team had assumed automation simply wasn’t available to them.
The company partnered with Squirrels.ai to deploy an AI AR Collections Specialist that runs on a weekly CSV export from their accounting system, bypassing the API limitation entirely. The agent now contacts every overdue customer by email, SMS, and phone on a structured cadence, holds real-time conversations, captures corrected email addresses on the first call, and updates each account with a clear status of paid, promised, or disputing. The single in-house AR specialist now has the equivalent of a full collections team working alongside them.
Client at a glance
Industry
Construction equipment rental (cranes and rigging)
Accounting system
Foundation construction accounting software (closed system, limited API)
Annual job volume
30 to 40 large jobs and roughly 2,000 smaller jobs per year
Use case deployed
AI AR Collections Specialist via weekly CSV upload
AR team
One dedicated AR specialist who also handles invoice generation
The challenge
The company’s AR operation was structurally constrained, not understaffed due to ill discipline. The team knew exactly what good collections looked like; they simply did not have the hours in the day to do it.
- One person, two jobs
A single AR specialist was responsible for both generating invoices and chasing overdue payments. With thousands of small-job invoices flowing through each year, invoice generation alone consumed most of their week. Collections follow-up that should have happened on day 30, day 45, and day 60 simply did not happen consistently, and aged invoices kept piling up.
- The closed accounting system
The company’s accounting software was a construction-industry-specific platform with very limited API access. Most automation vendors that had been evaluated previously required an open API and were therefore non-starters. The leadership team had effectively written off AR automation because of this constraint.
- The “never received it” problem
Roughly 90% of first collection calls were met with some version of “I never received the invoice.” Whether that was true or not, it stalled every conversation, required the AR specialist to verify the email on file, resend the invoice, and circle back later. Each one of those loops costs a day or more of payment timing.
Additional constraints
- Invoices were not currently delivered with a self-serve payment link, so any automation had to handle email re-sends gracefully.
- Some large customers needed to be excluded from automated outreach and kept on a personal touch.
- Cost discipline was important: the company wanted to start with the oldest, highest-impact invoices rather than blast every open invoice into the system.
The solution
Squirrels.ai deployed its AI AR Collections Specialist around the constraints of the existing tech stack rather than against them. The result was a working production system that did not require any change to the accounting platform.
Weekly CSV uploads instead of API integration
Rather than wait for an API that the vendor was not going to open, the team set up a simple weekly CSV export from the accounting system. The CSV carries everything Squirrels.ai needs: customer contact details, invoice number, due date, amount, and invoice description.
- No new infrastructure: the AR specialist exports a CSV once a week, the same way they already pull reports for their own use.
- Controlled scope: the team can filter the upload to a specific aging bucket, for example, 60 to 90 days overdue, to focus the AI on the highest-impact accounts and keep contact volume in check.
- Account-level exclusions: select large or relationship-sensitive accounts can simply be left off the upload so they continue to receive a human touch.
- Invoice descriptions included: because the CSV carries invoice line-item context, the agent can answer specific customer questions like “what was this charge for?” without having to escalate.
Multi-channel outreach sequences
- Coordinated cadence: the agent runs a configurable sequence of emails, SMS, and phone calls, escalating in tone as invoices age.
- Local-area caller ID: calls and texts come from numbers matched to the customer’s area code to improve pick-up and response rates.
- Real-time conversations: when a customer picks up, the AI handles the conversation naturally, references the specific invoice, and answers questions on the spot.
- Auto-stop on response: outreach pauses immediately when a customer pays, promises a date, or disputes the invoice.
- Company-owned email: emails are sent from the client’s own Google Workspace inbox so customers see a familiar sender.
Solving the “never received it” loop on the first call
FAQs were configured specifically for the most common objections. When a customer says they never received an invoice, the agent confirms the email address on file, captures a corrected address if needed, and offers to resend the invoice immediately, all in the same conversation. No more day-long round trips through the AR specialist.
Dashboards, recordings, and status tracking
- Every account is tracked with a live status: promised to pay, paid, disputing, no response, or contact-info issue.
- Call recordings, text transcripts, and email threads are all logged in one place for QA and listen-in monitoring.
- Channel-level analytics show which mix of email, text, and call is recovering the most dollars.
Results and impact
The company started with a two-week pilot focused on aged invoices in the small-jobs division. Within the pilot window, the team saw clear, measurable changes in how their AR motion ran.
What changed in the business
- Every overdue customer gets touched, every cycle. The high-volume small-jobs division now has consistent multi-touch follow-up that simply did not exist before, regardless of how busy the AR specialist’s week is.
- The AR specialist is freed for high-value work. Hours that used to be spent leaving voicemails and resending emails now go into invoice generation, exceptions handling, and the large, relationship-driven accounts that benefit from a personal touch.
- “Never received it” stops costing days. Corrected email addresses are captured on the first call, and the invoice is resent in the same conversation, instead of triggering a multi-day round trip.
- Costs are matched to value. By uploading only 60 to 90-day overdue invoices first, the team kept contact volume and platform costs in check while attacking the most painful aging bucket.
- Full visibility into every account. Promised-to-pay dates, disputes, and contact-info corrections all flow into one dashboard, with call recordings and transcripts on tap for QA.
- Closed software stopped being a blocker. A weekly CSV export turned out to be all the integration the team actually needed to put a production AI agent to work.
Why Squirrels.ai
Squirrels.ai deploys AI agents that handle calls, emails, and texts at scale, trained on a company’s own scripts, FAQs, and SOPs. For this company, the platform delivered:
- An AI AR Collections Specialist who worked around a closed accounting system using simple weekly CSV uploads.
- Real-time customer conversations that solved the “never received the invoice” problem on the first call instead of the third.
- Full control over which accounts get automated outreach and which keep a human touch, plus the ability to scope to specific aging buckets.
- Volume-based pricing and a structured two-week pilot so the team could validate ROI before scaling.