How Cityfurnish runs rentals end to end on an AI workforce

A relationship, not a transaction

Cityfurnish rents furniture, appliances, and fitness equipment on monthly subscriptions across India's major cities. Renting a sofa is not like buying one. The order is the beginning of a relationship that runs for months or years: signup and KYC, delivery and installation, an invoice every month, extensions, ownership transfers, and eventually a pickup.

Every one of those stages produces conversations. And half of them need Cityfurnish to reach out first instead of waiting for the customer to ask.

The work ecommerce never sees

A subscription rental business runs on operational conversations that a normal store never has. New signups stall at KYC. Payments fail at checkout and take a saved cart with them. Deliveries need scheduling calls, and then rescheduling calls. Invoices raise questions every month, and a missed one starts an overdue clock that can run for three hundred days.

Inbound was only half the job. The other half was outreach the team had to place manually, call by call and reminder by reminder, where the tone that works on day 3 of an overdue is already wrong by day 90.

Renewals, buybacks, and ownership transfers each carry their own paperwork and their own follow-ups. No human roster can hold all of it, every day, at the right moment, in the right tone.

A workforce, organised like a real team

Cityfurnish put an AI workforce on customer operations, with agents that each own a lane. Anjali runs inbound support across WhatsApp, Instagram, Messenger, and web chat, covering orders, KYC status, billing questions, refunds, and escalations that reach the right specialist team pre-briefed.

Eight voice agents make the outbound calls that used to be manual: Tanya welcomes new signups, Ayushi walks customers through KYC and pickup, Meera schedules deliveries, Sneha rescues failed orders, Muskan handles extensions and pickups, Dev runs retention, and Priya and Vikram manage collections from early overdue through 90-days-plus recovery. Nisha manages the community on Instagram and Facebook.

Underneath them, the workforce listens for lifecycle events, a cart abandoned, an invoice generated, a KYC rejected, a tenure ending, a payment overdue, and acts the moment one fires: the right message, email, or call, before the customer has to ask.

What the workforce gets done

In the last 30 days, the AI workforce handled more than 275,000 lifecycle touchpoints across WhatsApp, email, and voice. Subscription renewals alone drove 38,000+ of them, referral flows fired 18,000+ times after invoices and deliveries, and abandoned carts were re-engaged 21,000 times.

Overdue outreach now runs as a ladder from day 0 to day 300, seven bands deep, each with its own channel mix, and with voice joining at exactly the stages where a call outperforms a message.

• 275,000+ lifecycle touchpoints handled in 30 days
• Collections escalate on their own, day 0 to day 300
• Routine calls and reminders no longer placed by hand
• Humans see only the cases that need judgment, with the full story attached