We Know — Because We Built One
At Respage, we’ve already gone through this process. Twice. We built our own multifamily voice bot three years ago and then built it from the ground up less than a year later. Needless to say, it took a lot of people across multiple disciplines dedicated to this project for many months to get it done. And we’re still making tweaks. The reason? The technology around conversational AI was (and still is) evolving so quickly that we had to keep reinventing parts of the system to improve voice quality and optimize costs. AI services are not free. You’d be surprised how much the costs required to transcribe voice to text, generate a response, and translate it back to voice, not to mention store this data so you can listen to it or analyze it, add up.
This year, it was time to rebuild our website chatbot from scratch. And despite starting with a successful, proven foundation, re-engineering it took many months of work by a full team of developers and product experts. Even with the latest models, it’s not easy to get bots to answer questions without making stuff up sometimes. Not to mention a modern chatbot has to do more than just tell a prospect availability and help them schedule their tour. It needs to know how to cross-sell sister communities, add people to a waitlist, and get back to people as soon as those units are available.
That’s simply the reality. If your product depends on integrations and there are multiple suppliers who have built complex solutions, there probably isn’t ROI in building these systems unless you have enough units to support a multidisciplinary team of product managers, developers and QA.